Published April 29, 2016
o r i g i n a l r es e a r c h
Population Structure in the Model Grass Brachypodium
distachyon Is Highly Correlated with Flowering
Differences across Broad Geographic Areas
Ludmila Tyler, Scott J. Lee, Nelson D. Young, Gregory A. DeIulio,
Elena Benavente, Michael Reagon, Jessica Sysopha, Riccardo M. Baldini,
Angelo Troìa, Samuel P. Hazen, and Ana L. Caicedo*
Abstract
Core Ideas
The small, annual grass Brachypodium distachyon (L.) Beauv.,
a close relative of wheat (Triticum aestivum L.) and barley
(Hordeum vulgare L.), is a powerful model system for cereals and
bioenergy grasses. Genome-wide association studies (GWAS)
of natural variation can elucidate the genetic basis of complex
traits but have been so far limited in B. distachyon by the lack
of large numbers of well-characterized and sufficiently diverse
accessions. Here, we report on genotyping-by-sequencing (GBS)
of 84 B. distachyon, seven B. hybridum, and three B. stacei
accessions with diverse geographic origins including Albania,
Armenia, Georgia, Italy, Spain, and Turkey. Over 90,000
high-quality single-nucleotide polymorphisms (SNPs) distributed
across the Bd21 reference genome were identified. Our results
confirm the hybrid nature of the B. hybridum genome, which
appears as a mosaic of B. distachyon-like and B. stacei-like
sequences. Analysis of more than 50,000 SNPs for the B.
distachyon accessions revealed three distinct, genetically defined
populations. Surprisingly, these genomic profiles are associated
with differences in flowering time rather than with broad
geographic origin. High levels of differentiation in loci associated
with floral development support the differences in flowering
phenology between B. distachyon populations. Genome-wide
association studies combining genotypic and phenotypic data
also suggest the presence of one or more photoperiodism,
circadian clock, and vernalization genes in loci associated
with flowering time variation within B. distachyon populations.
Our characterization elucidates genes underlying population
differences, expands the germplasm resources available for
Brachypodium, and illustrates the feasibility and limitations of
GWAS in this model grass.
Published in Plant Genome
Volume 9. doi: 10.3835/plantgenome2015.08.0074
© Crop Science Society of America
5585 Guilford Rd., Madison, WI 53711 USA
This is an open access article distributed under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
the pl ant genome

ju ly 2016

vol . 9, no . 2 •
•
•
•
•
Genotyping diverse Brachypodium accessions
expands research tools for grasses.
The B. hybridum genome is a mosaic of B. distachyonand B. stacei-like sequences.
Three distinct, genetically defined populations of B.
distachyon were identified.
Flowering time, more than geography, distinguishes
B. distachyon populations.
Results support the feasibility of genome-wide
association studies in a model grass.
L. Tyler, Biochemistry and Molecular Biology Dep., Univ. of Massachusetts, Amherst, MA, 01003; S.J. Lee, Biology Dep., Univ. of Massachusetts, Amherst, MA, 01003 and Plant Biology Graduate Program, Univ. of Massachusetts, Amherst, MA, 01003; N.D. Young,
S.P. Hazen, and A.L. Caicedo, Biology Dep., Univ. of Massachusetts, Amherst, MA, 01003; G.A. DeIulio, Biochemistry and Molecular Biology Dep., Univ. of Massachusetts, Amherst, MA, 01003 and
Plant Biology Graduate Program, Univ. of Massachusetts, Amherst,
MA, 01003; E. Benavente, Dep. of Biotechnology, ETSIA, Technical
Univ. of Madrid, 28040 Madrid, Spain; M. Reagon, Dep. of Biology, The Ohio State Univ.–Lima, Lima, Ohio, 45804; J. Sysopha,
Biology Dep., Univ. of Massachusetts, Amherst, MA, 01003; R.M.
Baldini, Dep. of Biology, Univ. of Florence, 50121 Florence, Italy;
A. Troìa, Dipartimento STEBICEF, Sezione di Botanica ed Ecologia
vegetale, Università di Palermo, Palermo, 90133, Italy. Received
25 Aug. 2015. Accepted 29 Dec. 2015. *Corresponding author
([email protected]).
Abbreviations: F-statistic, FST; GBS, genotyping-by-sequencing; GLM,
general linear model; GO, gene ontology; GWAS, genome-wide
association studies; LD, linkage disequilibrium; MLM, mixed linear
model; MTA, marker–trait association; PCA, principal component
analysis; PCR, polymerase chain reaction; SNP, single nucleotide
polymorphism; SSR, simple-sequence repeat.
1
of
20
S
natural variation can elucidate the
genetic basis of complex, multigenic traits. In particular, next-generation genotyping has enabled GWAS,
which examine large, diverse populations for correlations
between alleles at individual markers and phenotypes
of interest. The GWAS approach has been successful in
a number of plant species, including rice (Oryza sativa
L.), maize (Zea mays L.), barley, the bioenergy grass
Miscanthus giganteus J. M. Greef & Deuter ex Hodk.
& Renvoize, and the dicotyledenous model Arabidopsis
thaliana (L.) Heynh. (Atwell et al., 2010; Huang et al.,
2010; Tian et al., 2011; Pasam et al., 2012; Slavov et al.,
2014). With the increasing demand for plants as sources
of food, feed, and fuel, there is a growing need for tools
to improve our understanding of agronomically important traits. Whether the investigations are performed in
crops or model species, the success of GWAS depends on
the availability of extensive, diverse, and well-characterized germplasm collections.
Here, we expand the germplasm resources available for the model grass B. distachyon. Brachypodium
distachyon was first proposed as a model research
organism nearly 15 yr ago (Draper et al., 2001) and has
rapidly developed into a powerful system with a wealth
of research tools (Brkljacic et al., 2011; Mur et al., 2011;
Girin et al., 2014). As a small, diploid annual with a compact, sequenced genome (International Brachypodium
Initiative, 2010) and amenability to both crossing and
genetic transformation (Vain et al., 2008; Vogel and Hill,
2008; Alves et al., 2009; Bragg et al., 2012), B. distachyon
has many of the advantages of the preeminent plant
model, A. thaliana (Koornneef and Meinke, 2010). However, unlike A. thaliana, B. distachyon is in the Pooideae
subfamily of the grasses, together with wheat, oat (Avena
sativa L.), and barley and is thus an especially valuable
model for the temperate cereals (Mochida and Shinozaki, 2013). Because it has the type II cell walls typical
of grasses, B. distachyon is also being used to investigate
feedstock properties for bioenergy production (Gomez
et al., 2008; Lee et al., 2012; Marriott et al., 2014; Tyler et
al., 2014). Despite all these advantages, germplasm collections of B. distachyon remain limited. Although the
native range of B. distachyon stretches across the Mediterranean region (Garvin et al., 2008; López-Alvarez et
al., 2015), material from only two countries, Spain and
Turkey, has been characterized in detail. This limitation
has hindered assessment of structure and other population parameters in the species, as well as development of
a diverse germplasm collection for GWAS.
Until recently, the diploid B. distachyon, with a haploid chromosome number of 5, was grouped in the same
species with morphologically similar Brachypodium
plants that have haploid chromosome numbers of 10
and 15. These latter two cytotypes are now considered
separate species: B. stacei (n = 10) and B. hybridum (n =
15) (Catalán et al., 2012). Fluorescence and genomic in
situ hybridization have elegantly demonstrated that B.
hybridum is an allotetraploid arising from interbreeding
tudying inherited
2
of
20
between diploid, B. distachyon-like and B. stacei-like
parents (Hasterok et al., 2004; Idziak et al., 2011; Catalán
et al., 2012). Flow cytometry has shown that the genome
of B. hybridum is approximately twice the size of the
B. stacei or B. distachyon genome (Catalán et al., 2012).
Furthermore, DNA barcoding and polymerase chain
reaction (PCR)-based analysis of simple-sequence repeats
(SSRs) have emerged as methods for Brachypodium species identification (Giraldo et al., 2012; López-Alvarez
et al., 2012). While B. stacei and B. hybridum are useful
for investigating speciation and ploidy, B. distachyon has
remained the premier model system within this genus,
and it is crucial to correctly distinguish the three species
when developing germplasm resources for GWAS.
In this study, we undertook the collection and characterization of 54 new Brachypodium accessions, together
with 40 previously described lines, originating from
the northern arc of the Mediterranean region and the
Middle East. Our collection includes novel material from
Spain, Italy, Albania, Georgia, and Armenia. We performed reduced-representation sequencing to generate a
large collection of polymorphic markers, which allowed
us to both differentiate between species and elucidate the
population structure within B. distachyon. Interestingly,
we identified flowering time as a major factor correlated
with population structure. Despite this correlation, and
the occurrence of strong structure in the species, we were
able to use flowering time as an example to illustrate the
feasibility of GWAS and limits to its resolution in our
extended germplasm panel.
Materials and Methods
Plant Materials and Growth Conditions
Between 2008 and 2012, we harvested seeds from individual plants in the field to develop five new lines from
Albania, four from Armenia, 10 from Italy, and 16 from
Spain. Seeds collected in the country of Georgia were
used to generate 18 new lines; however, because seeds
from more than one plant at each harvest site were
pooled, lines from the same site in Georgia could be
siblings. Also included in this study are two reference
B. hybridum lines, ABR100 and ABR113; one reference
B. stacei line, ABR114 (Draper et al., 2001; Hasterok et
al., 2004); and five previously characterized inbred B.
distachyon lines, Bd21, Bd21-3, Bd3-1, Bd30-1, and Bd1-1
(Vogel et al., 2006; Vogel and Hill, 2008; Schwartz et al.,
2010; Ream et al., 2014). Eighteen previously characterized inbred B. distachyon lines from Turkey (Filiz et al.,
2009; Vogel et al., 2009) and 14 additional accessions
from Spain (Giraldo et al., 2012) were included as well.
We derived a Ukrainian line by single-seed descent from
material deposited at the USDA National Plant Germplasm System under stock number PI 639818. Although
D. Garvin used the same stock to generate inbred line
Bd29-1 (Ream et al., 2014), the Ukrainian line used in
the current study (Ukr-Nvk1) and Bd29-1 are distinct.
A 95th line, BsUPM_AL1.1 (Giraldo et al., 2012), which
the pl ant genome

ju ly 2016

vol . 9, no . 2
had been identified as B. stacei, was also submitted for
GBS. However, because the GBS results and chromosome counting indicated that the submitted sample
was contaminated with DNA from a second accession,
BsUPM_AL1.1 was removed from subsequent analyses.
To maintain consistency with the previous nomenclature, while also promoting the clear identification of lines
in this study, three-letter prefixes indicating geographic
origin were combined with the accession name assigned
by the original collector. A three-letter country code (e.g.,
Ita for Italy)—and in some cases, a three-letter locality
code (e.g. Sic for Sicily)—was added to the beginning
of each accession name. For consistency, prefixes were
added even to the names of well-characterized reference
lines; for example, Irq-Bd21 indicates Bd21, the line that
originated in Iraq and was the source of the B. distachyon
reference genome sequence (International Brachypodium
Initiative, 2010). Supplemental Table S1 contains detailed
information about the final 94 lines used.
To synchronize germination and promote flowering,
seeds were cold treated before planting. Seeds were first
sterilized by removing the lemmas, washing the seeds for
4 min in a solution of 15% bleach plus 0.1% Triton-X 100
(Sigma-Aldrich), and rinsing the seeds twice with sterile
water. Seeds were placed on damp paper towels for stratification at 4C before sowing.
For DNA extraction and seed bulking, seeds were
sown in a 3:1 mixture of potting mix (#2, Conrad Fafard
Inc.) and Turface particles (Profile Products). Plants were
grown in a greenhouse with a 20-h photoperiod including supplemental lighting. Temperature varied somewhat
with the outside temperature at our location (Amherst,
MA, 42.392N, 72.525W). For phenotypic characterizations, seeds were sown in potting mix (Pro-Mix BX,
Hummert International), and plants were grown in an
environmentally controlled greenhouse under 16-h light
vs. 8-h dark conditions, with day vs. night temperatures
of 24 and 20C, respectively.
Seeds for all accessions that proved amenable to bulking are being deposited into the USDA National Plant
Germplasm System/Germplasm Resources Information
Network (NPGS/GRIN, http://www.ars-grin.gov/npgs/).
DNA Extraction and Genotypingby-Sequencing
DNA was extracted from leaf tissue of individual plants
or inbred lines using a DNeasy Plant Mini Kit (Qiagen)
with the following modifications: Buffer AP1 was heated
to 65C before use; a TE solution of 10 mM Tris, 0.1 mM
ethylenediamine tetraacetic acid (EDTA), pH 8 was used
for elution and was also heated to 65C.
DNA concentration was quantified using a Qubit 2.0
fluorometer and Quanti-iT dsDNA HS Assay Kit (Invitrogen). The quality of the genomic DNA was confirmed
by running samples on a 1% agarose gel for comparison
to a -DNA HindIII size standard (New England Biolabs)
and by digesting a subset of samples with HindIII (New
England Biolabs).
The DNA samples were submitted to the Cornell
University Institute for Genomic Diversity for GBS
(Ithaca, NY), as described previously (Elshire et al., 2011).
The GBS library with 96 unique barcodes represented
one blank and 95 genomic DNA samples (Supplemental
Table S1). Briefly, samples were digested with ApeKI,
ligated to barcoding and common adaptors, and amplified via PCR. The resulting fragments were tested for
size and sequenced in a single lane on an Illumina HiSeq
2000 using 100 nucleotide single-end reads. The mean
number of reads obtained per individual was 2.1 million.
Raw reads were submitted to the National Center for
Biotechnology Information Short Read Archive under
experiment SRX796779.
Filtering Sequencing Data for Single-Nucleotide
Polymorphism Calling and Quality Control
Single-nucleotide polymorphism calling and initial
filtering for the raw sequence reads were performed at
Cornell University, as described by Elshire et al. (2011),
using the GBS pipeline in TASSEL Version 3.0.138 (1
Nov. 2012). In summary, reads were checked for the presence of a barcode and at least 72 bases of good sequence,
trimmed to 64 bases, and aligned to the Bd21 v1.0 reference genome sequence, designated Bd21-1 (http://www.
brachypodium.org/gmod/genomic/accessions/1; file:
Bdistachyon.MIPS_1_2.fa.gz, downloaded 8 Nov. 2012)
(International Brachypodium Initiative, 2010). Sequence
tags mapping to multiple genomic loci were discarded.
The maximum number of good reads per lane was set to
4  108, the minimum sequence tag count to 5, and the
minimum minor allele frequency to 0.001. Heterozygotes were called when each allele was supported by the
minimum sequence tag count; loci with more than three
alleles were discarded. Minimum site coverage was set to
0.8, that is, SNP data had to be present for at least 80% of
the accessions for the SNP to be retained. This analysis
yielded a filtered set of 192,974 SNPs in HapMap format.
Next, we discarded SNPs that mapped to the multiple, small, extrachromosomal contigs included in the
Bd21 reference genome (International Brachypodium
Initiative, 2010). We further filtered SNPs to remove
those in or adjacent to mononucleotide repeats longer
than 5 bases and SNPs with missing data (N) for >10% of
the accessions. This additional filtering resulted in 99,872
high-quality SNPs for the full set of 94 Brachypodium
lines characterized by GBS (Supplemental Data S1). However, our later analyses indicated that 10 of 94 (11%) of
the accessions were B. stacei or B. hybridum (Supplemental Table S1). Repeating the filtering for the 84 putative B.
distachyon accessions yielded 54,392 high-quality SNPs
in a B. distachyon-only set (Supplemental Data S14).
Species Assignment via Sequencing
of the trnL-trnF Locus
Polymerase chain reaction primers flanking a trnL-trnF
intergenic region diagnostic for Brachypodium species
identification (López-Alvarez et al., 2012) were used to
t yler et al .: b . distachyon popu l ations di ffer in flowering 3
of
20
amplify part of the plastid trnLF locus from the same
DNA samples that were sent for GBS. As previously
described (Taberlet et al., 1991; López-Alvarez et al.,
2012), ‘c’ (5 CGAAATCGGTAGACGCTACG 3) and ‘f’
(5 ATTTGAACTGGTGACACGAG 3) universal primers were used for PCR. Amplification was performed
with ExTaq DNA polymerase (Clontech) under the following conditions: 98C for 30 s; 35 cycles of 98C for 10
s, 53C for 30 s, 72C for 1 min; and 72C for 10 min. The
forward ‘c’ primer was used for Sanger sequencing of the
PCR products. The sequences were manually trimmed
for quality and aligned in CLUSTAL W, with default settings, as implemented in MEGA5 (Tamura et al., 2011).
A 741-nucleotide region of the resulting alignment was
used to generate a neighbor-joining tree with bootstrap
analysis, p-distance substitution, and pairwise deletion
in MEGA5 (Felsenstein, 1985; Saitou and Nei, 1987; Nei
and Kumar, 2000; Tamura et al., 2011). In a few cases,
sequence at the trnLF locus was not determined, either
because of a lack of clear PCR or sequencing results
(Alb-AL1A, Alb-AL1D, and Geo-30i2) or because the
species assignment had been previously established (IrqBd3-1 and Por-ABR113). For comparison, an ABR113
sequence (GenBank: JN187658.1) determined by Catalán
et al. (2012) was included. New trnLF sequences were
submitted to Genbank (accessions KU163145-KU163233;
Supplemental Table S1).
Population Structure and Relatedness
The high-quality SNP sets were subjected to principal
component analysis (PCA) using SmartPCA in the
EIGENSOFT package (Patterson et al., 2006; Price et
al., 2006). The PCA used the full set of 99,872 or 54,392
SNPs for all-Brachypodium or B. distachyon-only analyses, respectively. The number of output principal components was set to 10 without removing outliers.
Subsets of the high-quality SNPs were analyzed
using version 2.3.3 of the STRUCTURE software
(Pritchard et al., 2000; Falush et al., 2003, 2007; Hubisz
et al., 2009) as installed on the Bioportal web service
at the University of Oslo, Norway, in July of 2013. Sites
with ambiguity codes were excluded from the analysis.
To limit problems associated with closely linked loci and
ease computational burden, SNPs 15 to 16 kb apart were
randomly selected, resulting in 10,169 or 9907 loci for
the all-Brachypodium or B. distachyon-only analyses,
respectively. Burn in was set to 100,000 and the number
of runs to 500,000. Because these Brachypodium species
are primarily selfing and cannot achieve Hardy–Weinberg equilibrium, ploidy was set to 1. An admixture
model was used to test the possibilities that K = 1 to K
= 15 populations with three replicates. The K method
(Evanno et al., 2005) was used to determine the number
of populations consistent with the SNP data.
Relationships among GBS haplotypes of B. distachyon and B. stacei were assessed using neighborjoining and maximum likelihood analyses. Heterozygous calls were converted to missing data to decrease
4
of
20
the impact of possible misalignment of paralogous
regions. Both neighbor-joining and maximum likelihood RAxML (Stamatakis, 2014) trees, based on 93,801
SNPs in total, were generated using iPlant apps (Goff
et al., 2011). The neighbor-joining tree was generated
using a Kimura two-parameter nucleotide substitution
model and 1000 bootstrap replicates to assess branch
support. The maximum likelihood tree was generated
using default settings and the general time-reversal plus
gamma (GTR + G) model. Unrooted trees were visualized with Dendroscope (Huson et al., 2007).
Calculating Timing of Population Divergence
We used DIYABC version 2.0 (Cornuet et al., 2014), which
applies approximate Bayesian computation (Beaumont et
al., 2002), to infer the time of divergence for the two main
B. distachyon populations. Sicilian accessions constituting PCA-defined Cluster 3A (see Results section) were
excluded from this analysis to reduce the effects of population substructure. We used a subset of 10,000 randomly
selected SNPs with a minor allele frequency cutoff of 5%.
We assumed a simple demographic model, which simulated a population that split at time t with no subsequent
gene flow between the resulting populations. Our model
included three parameters: the sizes of the two populations and time of divergence. The analysis using 200,000
simulations was performed on a 64-core computer server
at the University of Massachusetts–Amherst. Posterior
estimates of the three parameters were determined using
the logistic transformation option.
F-Statistic Analysis of Population Differentiation
We estimated SNP-specific FST between the two main B.
distachyon populations and searched for outliers using
the programs LOSITAN (Beaumont and Nichols, 1996;
Antao et al., 2008) and BayeScan 2.1 (Foll and Gaggiotti,
2008). Because of their status as a separate clade, and to
keep the analysis to two populations, Sicilian accessions
in PCA-defined Cluster 3A were excluded. Sites with a
minor allele frequency <5% were also removed. Heterozygous calls were recoded as missing data to decrease the
impact of possible misalignment of paralogous regions;
however, average FST values calculated with or without
heterozygous calls were very similar. In total, 26,897
SNPs were included. For LOSITAN, 50,000 simulations
were run using default parameters and both the neutralmean-FST and force-mean-FST options. Loci outside the
95% confidence interval were considered outliers. For
BayeScan, prior odds for the neutral model were set to
1, which assumes that the neutral model is as likely as a
model with selection. False discovery rate was set to 0.05,
and outlier detection was performed with an R function
provided by BayeScan. The SNPs identified as outliers or
as fixed by both programs were considered the best candidates for underlying population divergence. Putative
fixed SNPs further filtered out loci that had any (rare)
ambiguity code calls in the original dataset. We identified the B. distachyon gene closest to each candidate SNP
the pl ant genome

ju ly 2016

vol . 9, no . 2
and then identified rice homologs from the top hit using
BLASTx for protein similarity (Camacho et al., 2009).
These rice homologs were examined for gene ontology
(GO) term enrichment, as GO term annotation was
insufficient for analyses in B. distachyon. We used the
full MSU7.0 Oryza sativa nontransposable-element gene
identifier as background for term enrichment analysis
in agriGO (Du et al., 2010). Significance was evaluated
using a hypergeometric statistical test with a Hochberg
false discovery rate multiple comparison correction; the
minimum number of mapping entries was set to two.
Calculating Linkage Disequilibrium
Decay of linkage disequilibrium (LD) with distance was
calculated separately for the STRUCTURE-defined B.
distachyon Populations 1 (n = 57) and 2 (n = 23), excluding Sicilian accessions in PCA Cluster 3A to avoid overestimation of LD resulting from population substructure
as well as for the full set of B. distachyon (n = 84) using
TASSEL 4.0 (Bradbury et al., 2007). Ambiguity codes
were changed to N, and sites with more than 10% missing data were dropped. For the full set and Population
1, SNPs with a minor allele frequency of <5% were discarded. For Population 2, because of the smaller sample
size, a minor allele frequency cutoff of 8.5% was used.
Between 17,640 and 28,393 SNPs were used for the LD
calculations. The LD was evaluated with a sliding window 200 SNPs wide. A mean LD r2 value was taken from
all LD measures in each 1 kb bucket. Plots were generated using the ggplot2 package in R (Wickham, 2009; R
Development Core Team, 2014) and fitted with a generalized additive model curve. Further r2 means were calculated for other genomic distances (Table 1).
Flowering-Time Measurements
In Supplemental Data S10, the 63 accessions for which
flowering time was quantified are listed. Seed availability permitting, three or four seeds per accession per
treatment were surface sterilized as described above and
placed at 4C for 2 or 6 wk. Although some seeds—particularly those cold-treated for 6 wk—germinated in the
cold, seedling growth was minimal at 4C. Cold treatments were staggered so that seeds were all sown in the
greenhouse on the same day. Pots were arranged in a
randomized block design, and flats of pots were rotated
once a week. Flowering time was measured, from August
to January, as the number of days from sowing to initial
emergence of the inflorescence, that is, when the awn
of the top-most flower was first visible. This stage corresponds to the start of heading, stage 51 on the Biologische Bundesanstalt Bundessortenamt and Chemische
Industrie scale (Hong et al., 2011), and is consistent with
prior measurements of flowering time at stage 50 on the
Zadoks scale (Ream et al., 2014). Natural light intensity
varied over the course of our flowering-time experiment
at our greenhouse location in Amherst, MA (approximately 422330.9 N, 723155.1 W). Supplemental
lighting was used to maintain irradiance at or above 450
Table 1. Analysis of linkage disequilibrium for the full
set of 84 Brachypodium distachyon accessions and
subsets corresponding to Populations 1 and 2.
Full set
Range
Measures
kb
<1
101682
1–10
40004
10–20
38667
20–30
37868
30–50
72706
50–70
71140
70–100
106461
100–200
348034
200–300
343430
300–400
334274
400–500
335245
500–600
332148
500–700
330413
700–1000 958789
1000–5000 2164688
Population 1
r
2
0.141
0.327
0.293
0.273
0.264
0.248
0.233
0.213
0.191
0.178
0.168
0.164
0.159
0.155
0.168
Measures
r
93890
24077
23915
23152
43151
42269
62861
202529
195570
186478
186152
183407
182499
523570
1924432
0.126
0.409
0.360
0.326
0.301
0.267
0.236
0.194
0.151
0.128
0.113
0.104
0.097
0.089
0.078
2
Population 2
Measures
r2
90430
18151
16979
16366
31195
30496
45629
146346
142376
138778
138968
136280
136301
402062
1923235
0.175
0.408
0.350
0.331
0.305
0.284
0.270
0.250
0.227
0.217
0.203
0.197
0.188
0.184
0.188
W m−2 for 16 h d−1. For purposes of calculation, plants
that had not flowered by the end of the experiment at 149
d were assigned a flowering time of 150 d.
Genome-Wide Association Studies
Analysis of Flowering Time in Brachypodium
distachyon Accessions
Genome-wide association studies analysis was performed
using TASSEL 3.0 (Bradbury et al., 2007) for the 54,392
SNPS in the B. distachyon-only set, including individuals from Populations 1, 2, and 5 together with the
flowering-time data from plants that had received 6 wk
of vernalization as described above. We performed both
general linear model (GLM; Bradbury et al., 2007) and
mixed linear model (MLM; Zhang et al., 2010) analyses
via TASSEL. We used a 5% minor allele frequency cutoff
and verified that results were similar with a 10% cutoff.
A kinship matrix, generated within TASSEL, was added
as part of the MLM analysis to factor in relatedness
between accessions (Yu et al., 2006), and a Q + K model
was used. For the GLM, 1000 permutations were performed. The MLM with the preset optimum level of compression was used; variance components were estimated
after each marker. Quantile-quantile plots were compared to determine if MLM or GLM displayed a tighter
adherence to expected p-values.
To reduce Type I error rates that result from testing
with multiple SNPs, an adjusted significance threshold
was calculated. Because of our small population size,
we used a highly conservative method of dividing a traditional 0.05 p-value threshold by the total number of
SNPs tested. This adjusted significance threshold was
used as the genome-wide cutoff. The SNPs that crossed
t yler et al .: b . distachyon popu l ations di ffer in flowering 5
of
20
Fig. 1. Geographic distribution of Brachypodium accessions used in this study. Each dot represents an individual accession, with the
color of the outer ring representing inferred species (B. distachyon in dark blue, B. stacei in red, and B. hybridum in purple). The inner
pie chart represents the inferred ancestries of each accession as determined by STRUCTURE analysis (Population 1 in dark blue, Population 2 in cyan, Population 3 in red, Population 4 in yellow, and Population 5 in black). 1, the exact location of this collection site in
Iraq is unknown; 2, the collection site is Kalafabad, Iran (not shown). Detailed collection site information is listed in Supplemental Table
S1. Species assignments and STRUCTURE analysis are discussed in the main text.
this threshold were considered to represent significant
marker–trait associations (MTAs) (Supplemental Table
S4). To determine if candidate flowering-time regulators
were within or near the significant MTAs, we identified genes 250 kb upstream of the left border and 250 kb
downstream of the right border of each significant MTA
within the B. distachyon v1.0 reference genome. These
boundary genes were subsequently used to anchor MTA
intervals to the B. distachyon v3.1 gene annotation. We
explored the available annotation for all genes within
these regions including for the v3.1 annotation; resulting
candidates for regulating flowering time in B. distachyon
are reported in Supplemental Table S4.
Results
Diverse Germplasm Was Characterized
via Genotyping-by-Sequencing
Our study encompasses 94 Brachypodium accessions
from a broad geographic range and includes reference
lines for B. hybridum and B. stacei, as well as B. distachyon (Fig. 1; Supplemental Table S1). Many of the
samples were newly collected and developed for this
study and include countries for which Brachypodium
germplasm has not yet been characterized such as Italy,
Albania, Armenia, and Georgia. While some collection
sites are represented by a single accession, at other sites,
seeds were harvested from two to five different individuals. To maximize diversity, both newly harvested
germplasm and established inbred lines were included.
Names of accessions and collection sites are listed in
Supplemental Table S1.
Genotyping-by-sequencing (Elshire et al., 2011)
of these accessions yielded 99,872 high-quality SNPs
6
of
20
for the full set of 94 Brachypodium lines characterized
(Supplemental Data S1). These SNPs were fairly evenly
distributed across the five nuclear chromosomes of B.
distachyon, with one SNP approximately every 2.6 to
3.0 kb. However, as expected, the frequency of SNPs
was reduced near centromeres; ApeKI, the restriction
enzyme used to digest our genomic DNA samples, is partially methylation-sensitive and therefore unlikely to cut
within regions of repetitive DNA.
As a control, Bd21, the source of the fully sequenced
B. distachyon reference genome, was subjected to GBS
with the other 93 accessions. Only 111 SNPs were identified between our Bd21 sample and the reference genome,
suggesting a low likelihood of false-positive SNP calls.
The few SNPs that were identified in our GBS analysis of
Bd21 probably are due to a combination of sequencing
errors either in the reference or our data and remaining
trace heterozygosity in Bd21, as has been reported (Gordon et al., 2014). For the other accessions, GBS detected
from 3380 to 42,345 clear differences from the reference
(Supplemental Data S2).
The three known B. stacei lines had by far the highest
number of differences compared with Bd21, that is, more
than 40,000 SNPs per line. All other accessions had fewer
than 18,500 SNPs compared to the Bd21 reference (Supplemental Data S2). Pairwise comparisons of SNP calls
likewise emphasized the distinct nature of the B. stacei
lines ABR114, Ita-Sic-QVL1, and Ita-Sic-VMO1. Although
these lines were distinguished from each other by no more
than 1200 SNPs, upward of 35,000 SNPs often separated
B. stacei from other accessions. Exceptions to this general
trend included the reference B. hybridum lines ABR100
and ABR113, which differed from ABR114 by 7015 and
7161 SNPs respectively (Supplemental Data S3).
the pl ant genome

ju ly 2016

vol . 9, no . 2
Fig. 2. Levels of missing data and apparent heterozygosity in
genotyping-by-sequencing results for Brachypodium accessions.
The number of accessions with a given percentage of singlenucleotide polymorphisms represented by missing data (alleles
called as N’s, white bars) or nucleotide ambiguity codes (indicating two alleles, gray bars) is shown.
We also determined the proportion of missing SNP
information for each accession. For 85 out of 94 accessions, <10% of the SNPs corresponded to missing data,
testifying to the overall high quality of GBS (Fig. 2). Five
lines exhibited between 10 and 30% missing data, and
four lines had 40 to 50% missing data (Fig. 2). These last
four lines consisted of Bd3-1, an established B. distachyon
reference line from Iraq, and three B. stacei lines: ABR114,
Ita-Sic-QVL1, and Ita-Sic-VMO1. The appearance of
Bd3-1 as an outlier may indicate a technical issue with
this sample. The high proportion of missing data for the
B. stacei lines, however, probably is due to the fact that
SNPs were determined by mapping sequence reads to the
B. distachyon Bd21 genome. It is likely that many B. stacei
sequences were too divergent to meet the threshold for
mapping to B. distachyon loci and were therefore considered to be missing; alternatively, some B. distachyon loci
may be deleted in the B. stacei genome. In this context,
the thousands of SNPs differentiating B. stacei lines from
other accessions are even more striking (Supplemental
Data S2, S3); SNPs with missing nucleotide information were excluded from these totals, suggesting that the
GBS SNP calls represent a dramatic underestimate of the
actual, genome-wide sequence differences.
In addition, we examined the level of heterozygosity
in each accession (Fig. 2). The majority of accessions (87
out of 94, or 93%) had two alleles at <6% of the SNP loci;
in fact, 89% of the accessions had <2% heterozygosity.
This high level of homozygosity is consistent with previous characterizations of Turkish and Spanish germplasm
using SSR markers as well as an extremely high frequency of inbreeding reported for B. distachyon (Vogel
et al., 2009; Mur et al., 2011; Giraldo et al., 2012). In the
seven remaining accessions in our study, heterozygosity
was markedly higher, ranging from 21 to 38% (Fig. 2).
Both of the B. hybridum reference lines, ABR100 and
ABR114, had high levels of nucleotide ambiguity codes,
leading us to hypothesize that the other five lines in this
group (Alb-AL_1D and Spa-Nor-S1A, -S2C, -S7C, and
-S10C) might also be B. hybridum
The presence of both B. stacei-like and B. distachyonlike sequences for homeologous loci in the allotretraploid could result in the appearance of heterozygosity
for those loci. To confirm this possibility, we focused on
SNPs for which all three B. stacei lines lacked sequence
information. In this subset of 40,959 out of 99,872 highquality SNPs, ambiguity codes were found for no more
than 1.7% of the positions in any of the seven putative B.
hybridum. This observation is consistent with B. hybridum merely appearing to be highly heterozygous because
its genome is derived from divergent B. stacei-like and
B. distachyon-like parents. Together, the results on levels
of missing SNP data and heterozygosity (Fig. 2) were the
first indication that the GBS data clearly distinguished
between the three species (B. stacei, B. hybridum, and B.
distachyon) at a genome-wide level.
Brachypodium hybridum Is Intermediate
to Brachypodium stacei and Brachypodium
distachyon in Population Structure Analyses
To elucidate the population structure in our sample, we
investigated the complete set of 99,872 SNPs using the
distance-based method of PCA (Patterson et al., 2006;
Price et al., 2006). The 94 Brachypodium accessions
formed three clusters well separated from each other
along the first principal component, which explains the
largest percentage of the variation in the data (Fig. 3).
Because reference lines for B. stacei, B. hybridum, and
B. distachyon separated into PCA Clusters 1, 2, and
3, respectively, we hypothesized that species identity
explains the greatest amount of variation between the
accessions. Interestingly, the putative B. hybridum cluster
(Cluster 2) was centrally located between the B. stacei and
B. distachyon clusters along the first principal component
(Fig. 3). This result hints at the mixed parentage of the
allotetraploid. In addition, the 84 accessions in Cluster 3
formed three subgroups separating along the second principal component, which explains the next largest percentage of the variation (Fig. 3). While Cluster 3A consisted of
four Italian accessions, Clusters 3B and 3C were of mixed
geographic origin; Albania, Georgia, Italy, Spain, Turkey,
and Ukraine were represented in Cluster 3B and Armenia, Georgia, Iraq, Spain, and Turkey were represented in
Cluster 3C. The PCA thus suggests that the accessions in
Cluster 3 are all B. distachyon but that membership in the
subclusters is determined by a factor other than geography. Accessions comprising each cluster, with the corresponding values for the first 10 principal components, are
listed in Supplemental Data S4.
To further clarify the population structure, we analyzed the SNP data with STRUCTURE, a model-based
t yler et al .: b . distachyon popu l ations di ffer in flowering 7
of
20
Fig. 3. Principal component analysis clusters the three species separately, with Brachypodium hybridum intermediate to B. stacei and B.
distachyon. Principal component analysis was performed for 99,872 single-nucleotide polymorphisms using EIGENSOFT (Patterson et
al., 2006; Price et al., 2006). The values of principal component 1 and 2 (PC 1 and PC 2, explaining 29.9 and 10.0% of the variation
in the data set and represented by the x- and y-axes, respectively) were plotted for each accession. Each open diamond represents an
individual accession. Because of overlap, not all diamonds are clearly visible. Cluster 1 contains known B. stacei lines exclusively. Cluster 2 includes the reference B. hybridum lines ABR113 and ABR100. Cluster 3 contains reference B. distachyon lines in two of the three
subgroups: Bd1-1 in subgroup 3B, and Bd21, Bd21-3, and Bd3-1 in 3C. Subgroup 3A contains four Italian accessions.
clustering method (Pritchard et al., 2000; Falush et al.,
2003; Evanno et al., 2005; Falush et al., 2007; Hubisz et
al., 2009), and found a clear peak in K for K = 5 populations (Supplemental Data S5). A striking feature of the
inferred ancestry is that the putative B. stacei accessions
appeared to comprise a single population (Population
3, color-coded red in Fig. 4; Supplemental Data S6).
Moreover, this population was almost undetectable in
the inferred ancestries of B. distachyon individuals. In
B. distachyon, alleles from three populations— Population 1 (dark blue), Population 2 (cyan), or Population
5 (black)—tended to predominate (Fig. 4). Population
4 (yellow) contributed a relatively small proportion of
alleles to some accessions. The seven lines known or
hypothesized to be B. hybridum on the basis of the PCA
all exhibited mixed ancestries derived to varying degrees
from the B. stacei-like Population 3 plus B. distachyonlike populations (Fig. 4). Population 2, in particular, was
prominently represented in the inferred ancestries of all
the B. hybridum lines tested. This mosaic ancestry demonstrates, with the resolution achieved by thousands of
SNPs distributed across the chromosomes, that the B.
hybridum genome is a mixture of B. stacei-like and B.
distachyon-like sequences.
8
of
20
Genotyping-by-Sequencing Yields Unambiguous
Species Assignments Confirmed by Sequencing
of the DNA Barcoding Locus trnLF
To confirm the distinctions between species observed
with our GBS data, we employed the DNA barcoding
method established by López-Alvarez and colleagues for
the plastid trnLF locus (López-Alvarez et al., 2012). The
noncoding trnLF region, located between the chloroplast
trnL and trnF genes, is maternally inherited and distinct
between B. stacei and B. distachyon (López-Alvarez et al.,
2012). In most of the B. hybridum examined, the maternal parent appears to have been B. stacei-like (LópezAlvarez et al., 2012). Therefore, the presence of a B. stacei
trnLF sequence in a Brachypodium plant indicates that
the species is not B. distachyon, but could be either B. stacei or B. hybridum. For a small percentage of B. hybridum
individuals examined to date, the maternal parent seems
to have been B. distachyon-like (López-Alvarez et al.,
2012). Thus, identification of a B. distachyon-like trnLF
sequence could be consistent with a species assignment
of either B. distachyon or B. hybridum.
Figure 5 shows a neighbor-joining tree of trnLF
sequences for accessions analyzed by GBS including
reference B. stacei, B. hybridum, and B. distachyon. The
nucleotide sequences clearly formed two separate clades
each with 100% bootstrap support. The smaller clade
consisted of B. stacei-like sequences and included the
type B. stacei line ABR114 as well as the established B.
hybridum lines ABR100 and ABR113. Also in this small
clade were two Italian accessions (Ita-Sic-QVL1 and
the pl ant genome

ju ly 2016

vol . 9, no . 2
Fig. 4. STRUCTURE analysis reveals distinct genomic signatures for Brachypodium stacei, B. hybridum, and B. distachyon. The results
of a representative replicate for K = 5 populations are shown. Each of the 94 accessions is represented by a vertical bar. Color coding
along the y-axis indicates the proportion of the individual’s ancestry inferred to be derived from Population 1 (dark blue), 2 (cyan), 3
(red), 4 (yellow), or 5 (black). Accessions are ordered from left to right with B. stacei first, B. hybridum second, and then B. distachyon
accessions grouped by country of origin.
-VMO1) previously described as B. stacei (López-Alvarez
et al., 2012) and four Spanish lines (Fig. 5). These four
Spanish accessions—like the B. hybridum reference
lines—were in PCA Cluster 2; their membership in the
smaller trnLF clade is consistent with a B. stacei-like
maternal parent for the hybrid species (Fig. 5). The larger
clade in the tree consisted of B. distachyon-like sequences
and included the reference B. distachyon lines Bd21, Bd11, and Bd21-3 in addition to the remaining accessions for
which clear trnLF sequence was available (Fig. 5). Overall, the trnLF tree (Fig. 5; Supplemental Table S1) supported the idea that accessions separated along the first
principal component (Fig. 3) based on species and that
population structure captures species designations (Fig.
4). When other researchers had made previous species
assignments, our combined PCA, trnLF, and STRUCTURE results agreed with the earlier assignments (Filiz
et al., 2009; Vogel et al., 2009; López-Alvarez et al., 2012).
Thus, our analysis of thousands of SNPs not only reliably
distinguished between species but also provided a deeper
understanding of the genomic composition of individuals within these species.
Geographic Origin Does Not Fully Explain
the Population Structure Observed within
Brachypodium distachyon
Within the 84 B. distachyon accessions, the three subgroupings revealed by PCA and STRUCTURE did not
correlate broadly with geographic origin (Supplemental
Data S4, S6). The divergent genomic signatures are particularly apparent in Fig. 4 (see also Fig. 1), in which the
B. distachyon accessions are ordered by country of origin
roughly from west to east. Notably, representatives of
STRUCTURE Populations 1 and 2 are present in both
the western and eastern edges of the geographic distribution (Fig. 4), although Population 1 does predominate in
Spain. Additionally, in several instances, accessions harvested from the same location had dramatically different
inferred ancestries. For example, Ita-Sic-SLZ2 and-SLZ4
originated from different individuals at the same site in
Schilizzi near Palermo, Sicily (Supplemental Table S1);
however, STRUCTURE Populations 2 and 5 represented
91 and 92% of the inferred ancestry in Ita-Sic-SLZ2 and
Ita-Sic-SLZ4, respectively. Similarly, Geo-G31i2, -G31i4,
and -G31i6 all originated from a single site in southeastern Georgia but exhibited alleles primarily from STRUCTURE Population 2 in Geo-G31i2 and Population 1 in
Geo-G31i4 and -G31i6. Other accessions with the same
geographic origin but contrasting genomic profiles
t yler et al .: b . distachyon popu l ations di ffer in flowering 9
of
20
Fig. 5. Sequences for the plastid trnLF locus distinguish Brachypodium distachyon from B. stacei and B. hybridum. This neighbor-joining
tree—generated in MEGA5 (Tamura et al., 2011)—is based on nucleotide sequences from the region between the chloroplast trnL and
trnF genes for accessions analyzed by genotyping-by-sequencing. The tree is a consensus from 500 bootstrap replicates. Numbers
indicate bootstrap support, and only branches with >50% support are shown. Blue lines indicate B. distachyon-like sequences, while
red lines indicate B. stacei-like sequences. Blue squares mark the reference B. distachyon lines Bd21, Bd1-1, and Bd21-3; purple circles
mark the reference B. hybridum lines ABR113 and ABR100; and a red triangle marks the reference B. stacei line ABR114.
included Geo-G32i2, -G32i4, -G32i6, and -G32i7; GeoG34i2, -G34i3, and -G34i6; and Spa-Nor-S6B, -S6D, and
-S6E (Fig. 4; Supplemental Data S6).
A neighbor-joining phylogenetic tree constructed for
B. stacei and B. distachyon with the full SNP set revealed
local geographic structure within major B. distachyon
groups, but no geographic structure between major
groups (Supplemental Fig. S1). The two major clades were
monophyletic with high support and were consistent
with the division of most B. distachyon accessions into
PCA Clusters 3B and 3C or STRUCTURE Populations 2
and 1. Tree analysis further supported the distinction of
10
of
20
PCA Cluster 3A or STRUCTURE Population 5 as a separate clade of Italian accessions more closely related to
STRUCTURE Population 2 (Supplemental Fig. S1). Major
clades tended to represent multiple geographic origins,
though some of the subclades represented single regions
such as Albania, Italy, northern Spain, and Turkey,
indicating local relatedness among many B. distachyon
samples (Supplemental Fig. S1).
To investigate whether species identity was obscuring the true population structure within B. distachyon,
we excluded B. stacei and B. hybridum accessions and
repeated the PCA using the 54,392 SNPs polymorphic
the pl ant genome

ju ly 2016

vol . 9, no . 2
for B. distachyon. This PCA separated the B. distachyon
accessions into three well-defined groups for the first
two principal components (Supplemental Data S7), consistent with the all-Brachypodium analysis: the four Italian accessions Ita-Sic-CSR6, -CSR7, -LPA3, and -SLZ4
formed a distinct cluster (3A); 23 accessions from six
countries comprised a second cluster (3B); and 57 B.
distachyon accessions from five countries comprised the
third cluster (3C; Supplemental Data S7). Only along the
third vs. fifth principal components did accessions begin
to form clusters consistent with geographic origin. Turkish accessions separated from Spanish ones along the
third principal component, while Albanian and Georgian accessions separated along the fifth principal component (Supplemental Data S7). Thus, although evidence
of relatedness as a result of similar geographic origin was
apparent, geography was not the primary, defining feature of the population structure.
We also analyzed 9907 SNPs selected from the B.
distachyon-only set with STRUCTURE. In this case, K
exhibited a peak for two populations, and PCA-defined
subgroups 3A and 3B were combined, consistent with
their sister-clade designation in the phylogenetic trees.
However, subgroups 3A and 3B remained clearly separated from 3C and included accessions from multiple
countries as above (Supplemental Data S8, S9).
Flowering Time Differences Are Largely
Consistent with the Brachypodium distachyon
Population Structure
Observing B. distachyon clusters of mixed geographic
origin prompted us to look for another factor that might
more fully explain population structure. Although limited phenotypic information was available for most of the
accessions, the reference Bd lines were known to have
contrasting flowering times. Notably, Bd21, Bd21-3, and
Bd3-1—all of which derived most of their ancestry from
the all-Brachypodium STRUCTURE Population 1 (dark
blue)—are consistently early flowering, while Bd1-1,
assigned to Population 2 (cyan), is late-flowering (Vogel
et al., 2009; Schwartz et al., 2010; Ream et al., 2014). In
addition, whereas many Turkish accessions, including
Adi-4 and BdTR9m, flower relatively early, a few, such
as Tek-9, are extremely late flowering (Filiz et al., 2009;
Vogel et al., 2009). Bd1-1 and Tek-9 were the only Turkish
lines in our collection assigned primarily to Population
2 (cyan) (Fig. 4). Interestingly, our line Ukr-Nvk1, with
alleles predominantly from Population 2, originated
from the same seed stock as Bd29-1, which has been
characterized as delayed flowering similar to Bd1-1
(Ream et al., 2014). Moreover, as plants were growing for
tissue and seed harvesting, we noticed that B. distachyon
accessions with either Population 2 (cyan) or Population 5 (black) comprising most of the inferred ancestry
tended to flower late. To determine whether this trend
held, we quantified the flowering time for three-fourths
of the B. distachyon accessions.
At the time of testing, sufficient seed was available for
48 of 57 (84%) of the Population 1 accessions (dark blue),
12 of 23 (52%) of the Population 2 accessions (cyan), and
three of four (75%) of the Population 3 accessions (black).
Three or four seeds for each accession were cold treated
for 2 or 6 wk and then sown in soil and grown with a photoperiod of 16 h. Figure 6 shows the resulting flowering
times. Supplemental Data S10 includes the full data set.
As has been reported previously (Vogel et al., 2009;
Schwartz et al., 2010; Ream et al., 2014; Tyler et al., 2014),
extended cold treatment tended to trigger earlier flowering in most accessions (Fig. 6A, 6C; Supplemental Data
S10). For example, Bd21-3 cold treated for 6 wk flowered
31.2 d earlier than with the 2-wk cold treatment. Dramatic variation in flowering times between accessions
was also observed (Fig. 6B, 6C), with Bd3-1 flowering at
18 d and several late-flowering accessions, such as Ita-SicSLZ2 and Tek-9, remaining vegetative over the course of
the experiment even after 6 wk of cold treatment (Fig. 6;
Supplemental Data S10). Because a day length of 20 h can
further promote flowering (Ream et al., 2014), the 16-h
photoperiod used for this study likely accentuated differences in vernalization responsiveness.
To determine whether flowering time was correlated
with the observed population structure, we averaged
the flowering times of accessions in PCA Cluster 3C–
STRUCTURE Population 1 and compared them with
flowering times of accessions in Cluster 3B–Population 2
for the 6-wk cold treatment (Fig. 6D). Cluster 3A–Population 5 was omitted because of its small sample size.
Strikingly, although the variability within each cluster
was high, Population 1 accessions had an average flowering time of 58.2 d compared with 122.2 d for Population
2 accessions (Fig. 6B). A single-factor analysis of variance
(ANOVA) indicated that this difference between populations was statistically significant at P = 7.63  10−6. However, outliers occurred; for example, the Armenian accessions Arm-3A, -3G, and -3i as well as the Georgian accession Geo-G34i6, all of which were assigned to Population
1 (blue), required 130 d or more to flower (Supplemental
Data S10). The earliest-flowering Population 2 accession
was Spa-Nor-S6D. With an average flowering time of 27.7
d (2.5 d SD), Spa-Nor-S6D flowered even earlier than
many Population 1 accessions. Interestingly, two other
accessions, Spa-Nor-S6B and -S6E, which originated
from the same site as Spa-Nor-S6D, were relatively early
flowering and had Population 1-type genomic signatures.
In summary, in spite of the intriguing exceptions, flowering time differences are a major distinction between the
main B. distachyon populations.
Many Single-Nucleotide Polymorphism
Are Highly Differentiated or Fixed between
Brachypodium distachyon Populations 1 and 2
We examined the degree of divergence between the two
principal B. distachyon populations with F-statistic (FST)
analyses using the programs LOSITAN and BayeScan
t yler et al .: b . distachyon popu l ations di ffer in flowering 11
of
20
Fig. 6. Vernalization responsiveness and flowering time vary between accessions. Flowering time was measured after seeds had been
cold treated for 2 or 6 wk. (a) Plants of the Population 1 accession Spa-Nor-S11B were photographed after 8 wk of growth; the plants
on the left and right experienced 6 and 2 wk of cold treatment, respectively, before sowing. (b) Plants that had been cold treated for
6 wk as seeds were photographed after 9 wk of growth; Bd3-1 is from Population 1 (blue), Alb-AL1A from Population 2 (cyan), and
Ita-Sic-LPA3 from Population 5 (black). The scale bars in (a) and (b) represent 10 cm. (c) Flowering time is shown as the average number of days required for flowering for two to four plants per accession per treatment. Plants that had not flowered by the end of the
experiment were assigned a flowering time of 150 d. Most accessions are represented by black lines; select accessions are highlighted
as follows: Irq-Bd3–1, solid, dark blue line; Irq-Bd21, dotted, blue line; Irq-Bd21-3, dashed, blue line; Tur-Bd1-1, solid, cyan line; and
Tur-Tek-9, dashed, cyan line. (d) On average, after 6 wk of cold treatment, Population 1 accessions flowered earlier than Population 2
accessions. The black triangles indicate the average flowering time in days. The shaded boxes represent the second and third quartiles,
while the whiskers indicate the maximum and minimum flowering times for accessions in the two populations. The median is represented by a horizontal black line across the shaded box for Population 1 and is at 150 d for Population 2. Sample sizes are 48 and 10
accessions for Populations 1 and 2, respectively.
12
of
20
the pl ant genome

ju ly 2016

vol . 9, no . 2
(Beaumont and Nichols, 1996; Antao et al., 2008; Foll
and Gaggiotti, 2008). The estimated FST for the dataset
was high for both programs (0.51 and 0.62, respectively),
indicating extensive differentiation across the genome
between populations of B. distachyon. In total, LOSITAN
identified 632 loci as fixed (FST = 1) and 2772 loci as high
FST outliers; BayeScan identified 686 loci, including fixed
loci, as outliers (Supplemental Data S11, S12). We focused
on the 218 fixed loci between Populations 1 and 2, which
occur on all five B. distachyon chromosomes. In addition,
14 SNPs on chromosomes 1, 2, and 3 were identified by
both programs as highly differentiated (Supplemental
Data S13). Besides indicating the types of loci that differentiate the two populations, this set of SNPs could be
useful for genotyping new germplasm to determine likely
membership in Population 1 vs. 2.
We determined the identity of the B. distachyon
gene nearest each of these highly divergent or fixed SNPs
and then identified the closest rice homologous gene
(Supplemental Table S2). Gene ontology term analysis of
the rice orthologs revealed 20 biological process terms
significantly overrepresented in our dataset (Table 2;
Supplemental Table S3; Supplemental Fig. S2). Remarkably, the majority of these terms are related to reproduction and developmental processes including flower
development (Table 2; Supplemental Fig. S2). Inspection
of the individual annotations for the rice and closest A.
thaliana orthologs likewise revealed various genes associated with flowering, organ development, and meristem
maintenance including a DELLA-domain GRAS transcription factor gene, a gibberellin-receptor-like GID1L2
gene, AINTEGUMENTA, APETALA1, FERTILIZATIONINDEPENDENT ENDOSPERM, AUXIN RESPONSE
FACTOR10, BRASSINOSTEROID INSENTITIVE1, HECATE, LATERAL SUPPRESOR, PERIANTHIA, SEPALLATA1, REPLUMLESS, SQUAMOSA PROMOTER BINDING PROTEIN-LIKE9, LOST MERISTEMS3, FILAMENTOUS FLOWER, and TARGET OF MONOPTEROS6
(Ikeda et al., 2001; Greb et al., 2003; Malcomber and
Kellogg, 2005; Nole-Wilson and Krizek, 2006; Gremski
et al., 2007; Schwarz et al., 2008; Mosquna et al., 2009;
Ding and Friml, 2010; Li et al., 2010; Schlereth et al.,
2010; Schulze et al., 2010; Maier et al., 2011; Jiang et al.,
2014; Pabón-Mora et al., 2014) (Supplemental Table S2).
These results suggest that genes involved in flowering
and developmental programming are among the most
highly differentiated between the two B. distachyon populations. It is possible that differences in at least some of
these genes may play a role in the flowering time differences observed between populations under our common
garden conditions.
Given the high levels of differentiation between
B. distachyon populations, we attempted to determine
when they had diverged from each other using DIYABC
(Cornuet et al., 2008, 2014). The median of the posterior
distribution for the estimated time of divergence was
5920 generations (95% confidence interval: 1290–9620)
(Supplemental Fig. S3). Because B. distachyon is annual,
with likely one generation per year, this result indicates
that divergence between B. distachyon populations
occurred <10,000 yr ago. In contrast, the B. stacei and B.
distachyon lineages split an estimated 9.9 and 4.3 million
yr ago from the common ancestor, respectively (LópezAlvarez et al., 2012). The relatively recent divergence
observed here for B. distachyon populations, after the
end of Pleistocene glaciations, suggests that differences
in flowering phenology evolved rapidly among B. distachyon populations.
Linkage Disequilibrium Declines More Quickly
in Population 1 than Population 2
Having detected broad flowering time differences and
highly differentiated SNPs between Populations 1 and
2, we next examined LD to assess the resolution of association mapping in these populations of B. distachyon.
Linkage disequilibrium describes the degree of association between alleles at different loci; rapid decay of LD
within a short genomic distance is favorable for GWAS,
as it decreases the number of markers associated with a
phenotype of interest and allows much finer resolution
when mapping. We examined the decay of LD with distance in Population 1, Population 2, and the full set of B.
distachyon accessions. Figure 7 shows how LD measures
change with distance; Table 1 summarizes the results of
this analysis in bins ranging from 10 to 5000 kb. Linkage disequilibrium decay is not particularly rapid in B.
distachyon; in the full set, average r2 levels of 0.1 are not
reached by 5 Mb. This is in contrast to maize, where r2
levels of 0.1 are reached on average by 2 to 5 kb (Yan et
al., 2009), or A. thaliana, where levels of 0.2 are reached
by about 10 kb (Kim et al., 2007). The highly selfing
nature of B. distachyon (Vogel et al., 2009) can lead to
high levels of LD; however, LD levels in B. distachyon
can also be affected by the strong population structure
within the species. When Populations 1 and 2 are examined separately, LD decay occurs more rapidly. In particular, LD decay is most rapid in Population 1, where r2
levels close to 0.1 are reached by ~500 kb, and half-maximum values are reached by 100 to 200 kb, similar to what
is observed in some varieties of cultivated rice (Mather
et al., 2007; Xu et al., 2012). In contrast, LD decay is quite
slow in Population 2, with levels of r2 of 0.1 again not
reached by 5 Mb and half-maximum values reached by
400 to 500 kb. Extensive LD within this population may
be due to the moderate sample size (n = 23). Together,
these results suggest that GWAS will have greater resolution if mapping is performed within Population 1. Nevertheless, increased mapping resolution could be achieved
if the LD rate of decay were increased, perhaps through
sampling of more accessions, even in this population.
t yler et al .: b . distachyon popu l ations di ffer in flowering 13
of
20
Fig. 7. Linkage disequilibrium decays more quickly in Brachypodium distachyon Population 1 than Population 2. Linkage disequilibrium
(LD) plots for the complete set of 84 B. distachyon accessions, 57 Population 1 accessions, and 23 Population 2 accessions are shown
in (a), (b), and (c), respectively. Linkage disequilibrium was not analyzed separately for the four Population 5 accessions because of the
small sample size. Pairwise LD values (r2) are plotted against physical distance. Trend lines represent plot data fitted with a generalized
additive model curve.
Table 2. Significantly overrepresented biological process gene ontology (GO) terms among genes near
highly divergent SNP markers.
GO term
Description
GO:0006807 Nitrogen compound metabolic process
GO:0022414
Reproductive process
GO:0000003
Reproduction
GO:0006139 Nucleobase, nucleoside, nucleotide
and nucleic acid metabolic process
GO:0048856 Anatomical structure development
GO:0009058
Biosynthetic process
GO:0009908
Flower development
GO:0048608 Reproductive structure development
GO:0003006 Reproductive developmental process
GO:0030154
Cell differentiation
GO:0048869 Cellular developmental process
GO:0032501 Multicellular organismal process
GO:0009653 Anatomical structure morphogenesis
GO:0007275 Multicellular organismal development
GO:0009791
Post-embryonic development
GO:0016049
Cell growth
GO:0090066
Regulation of anatomical
structure size
GO:0032502
Developmental process
GO:0008361
Regulation of cell size
Genome
False
Input total
discovery
count count p-value rate
43
15
22
43
3898
732
1449
3898
0.00046
0.00013
0.00028
0.00046
0.0077
0.0077
0.0077
0.0077
19
53
10
10
10
10
10
29
13
28
18
8
8
1250
5311
469
474
474
534
534
2636
886
2591
1441
465
465
0.00074
0.00093
0.0013
0.0015
0.0015
0.0034
0.0034
0.0048
0.0068
0.0071
0.0079
0.013
0.013
0.0099
0.01
0.011
0.011
0.011
0.021
0.021
0.027
0.034
0.034
0.035
0.045
0.045
29
8
2806
465
0.011
0.013
0.045
0.045
Genome-Wide Association Mapping Reveals
Significant Associations for Flowering Time after
Six Weeks of Vernalization
For many of the accessions tested, vernalization of 2 wk
was insufficient to trigger flowering before the experiment ended (Supplemental Data S10). Therefore, to test
the performance of GWAS in B. distachyon, we searched
14
of
20
for significant MTAs between our high-density SNP data
and time of flowering for 60 accessions after 6 wk of vernalization. We initially attempted GWAS using Population 1 only because of the relatively low LD within this
group and to limit the effects of population structure.
After correcting for multiple tests, no significant MTAs
were identified within this small sample of accessions.
Accordingly, we used the complete set of phenotyped
accessions despite the risk of false negatives as a result
of an association of flowering time with population
structure (Brachi et al., 2011). Comparison of quantile–
quantile plots demonstrated that the MLM displayed a
tighter adherence to expected P-values than the GLM,
though even with the MLM, P-values were still somewhat inflated (Supplemental Fig. S4). Focusing on results
from the MLM analysis, we used a conservative adjusted
significance threshold by dividing the 0.5 significance
threshold by the number of SNPs tested, resulting in a
genome-wide significance cutoff of 1.62  10−6. We identified nine distinct regions of MTAs comprised of one or
more significantly associated SNPs (Fig. 8). Four of the
nine MTAs consisted of a single SNP, while the remaining MTAs contained multiple SNPs (Supplemental Table
S4). Clear phenotypic differences were observed between
individuals carrying alternative alleles at each MTA
(Supplemental Fig. S5). Thus, these nine MTAs represent
genomic loci that may significantly influence flowering
time following 6 wk of vernalization.
The genetics of flowering time has been extensively
studied in numerous species, including grasses, and 163
predicted flowering-time genes have been described in
B. distachyon (Higgins et al., 2010; Ream et al., 2014).
We therefore compared locations of these B. distachyon
genes with MTAs (Fig. 8; Supplemental Table S4). As a
conservative cutoff for distance, based on the decay of
LD in Population 1, we chose 250 kb. In five cases, candidate genes lay within 250 kb of a significant MTA or
the pl ant genome

ju ly 2016

vol . 9, no . 2
Fig. 8. Manhattan plot showing single-nucleotide polymorphism marker associations with flowering time following 6 wk of vernalization
across the five chromosomes of Brachypodium distachyon. The chromosomal locations of B. distachyon homologs of previously characterized flowering time, circadian rhythm, and photoperiod genes are indicated. A horizontal gray dashed line represents the adjusted
genome-wide significance threshold.
associated interval. For example, BdCONSTANS1 (CO1),
a homolog of A. thaliana CONSTANS, is 27 kb outside
the MTA interval on chromosome 1. In A. thaliana, CO
expression in the dark activates FLOWERING LOCUS
T (FT), which encodes the mobile florigen signal that
initiates the floral transition (Kardailsky et al., 1999;
Samach et al., 2000; Jaeger and Wigge, 2007), and a rice
gene closely related to CO, Hd1/photoperiod sensitive1,
influences flowering time differentially under short and
long days (Yano et al., 2000). A B. distachyon homolog
of the A. thaliana circadian clock gene EARLY FLOWERING3 (ELF3) lies within the boundaries of the MTA
interval on chromosome 2. This gene was originally
characterized in a screen for early flowering A. thaliana
mutants (Zagotta et al., 1992). In barley, mutations in
Praematurum-a, the barley homolog of ELF3, result in
similar early flowering phenotypes (Zakhrabekova et al.,
2012). On chromosome 3, one of the two MTAs is 150
kb from a gene orthologous to A. thaliana GROWTH
FACTOR REGULATOR 7 (GFR7). Although GFR genes
have not been shown to directly impact flowering time,
they have been implicated in floral organ differentiation
and meristem patterning (Pajoro et al., 2014). Two of the
four MTAs on chromosome 4 are located <150 kb from
BdPSEUDO-RESPONSE REGULATOR59 (BdPRR59) and
BdPRR95, homologs of the A. thaliana circadian clock
genes PRR5 and PRR9 (Nakamichi et al., 2005). While
the role of PRRs in grasses is less well understood than
in A. thaliana, in wheat, a 2 kb deletion upstream of the
PRR family member Ppd-D1 resulted in early flowering
in both short and long days (Beales et al., 2007).
The remaining four MTAs—one on chromosome 3,
two on chromosome 4, and one on chromosome 5—are
more than 250 kb from obvious a priori candidates.
These MTAs may correspond to novel, previously undetected, flowering-time-regulatory genes, or, given the
slow LD decay in our B. distachyon sample, may be associated with more distant candidate genes.
Because the B. distachyon genome is compact, many
potential flowering-time genes have been described, and
we used a relatively large (250 kb) cutoff for loci associated with an MTA, we considered whether obtaining five
t yler et al .: b . distachyon popu l ations di ffer in flowering 15
of
20
MTAs close to flowering-time candidate genes could have
occurred by chance. In a random sample of 10,000 of our
GBS markers, 1738 markers lie 149 kb or less from a candidate flowering-time gene listed in Higgins et al., 2010 and
Ream et al., 2014. Also, 149 kb represents the greatest distance we observed for the five MTAs we considered to be
close to flowering-time candidates. This suggests there is a
17.4% chance of any MTA falling close to a flowering-time
gene whether or not there is a true association. In a sample
of nine MTAs, we thus would have expected 1.56 to be
within 149 kb of a flowering-time gene by chance. With
five MTAs at distances of 149 kb from a flowering-time
candidate, our sample does seem enriched for described
flowering-time loci. Though more work is needed to confirm involvement of these genes in flowering-time variation in B. distachyon, our GWAS analysis has provided
some fruitful candidates for exploration.
Discussion
By identifying SNPs adjacent to restriction enzyme sites,
GBS captured a fraction of the genetic diversity in this
germplasm collection. High-coverage, whole-genome
resequencing would undoubtedly uncover an even
greater richness of genetic variation including insertions,
deletions, and rearrangements. For instance, resequencing of six core B. distachyon lines revealed upward of 1.6
million SNPs between the late-flowering line Bd1-1 and
the early flowering line Bd21 (Gordon et al., 2014). In
comparison, our GBS analyses detected approximately
12,000 SNPs between Bd1-1 and Bd21. The cost-effectiveness of reduced-representation sequencing, however,
allowed us to survey genetic diversity across a greater
number of accessions. The SNP information yielded by
this broad approach made it possible to distinguish the
genomic signatures of three Brachypodium species, characterize population structure within B. distachyon across
its native range, and identify candidate loci potentially
controlling flowering time.
In multiple analyses (PCA, STRUCTURE results,
and phylogenetic trees), B. stacei, B. hybridum, and B.
distachyon were clearly distinct (Fig. 3, 4; Supplemental
Fig. S1). Because SNP calling involved mapping sequences
to the B. distachyon reference genome, this distinction
was also evident on examining levels of missing SNP
data and heterozygosity (Fig. 2). A practical application of
these findings is the identification of numerous candidate
SNPs for use as markers in PCR-based species assignments; examination of those SNP loci for which sequence
information was present for all three B. stacei accessions
showed that, in many cases, the B. stacei and B. distachyon
accessions carried contrasting nucleotides, while the seven
B. hybridum accessions uniformly showed evidence of
both B. stacei-like and B. distachyon-like alleles (Supplemental Data S1, S14). Although additional B. stacei and B.
hybridum accessions must be tested to confirm the broad
utility of each marker, subsets drawn from each chromosome could be used as cleaved, amplified polymorphic
16
of
20
sequence markers (Konieczny and Ausubel, 1993) to rapidly assign species membership.
This study elucidates the differences between species
across the nuclear genome at an average resolution of <5
kb. Our work thus reinforces and refines the three-species
concept first revealed by fluorescence in situ hybridization
and examination of organellar and nuclear loci (Hasterok
et al., 2004; Idziak et al., 2011; Catalán et al., 2012; Giraldo
et al., 2012; López-Alvarez et al., 2012). The PCA, summarizing information from nearly 100,000 SNPs, placed
B. hybridum intermediate to B. stacei and B. distachyon
along the first principal component (Fig. 3); thus, species
identity is, genome-wide, the primary distinguishing factor for the 94 accessions in our collection. Additionally,
STRUCTURE analysis showed the extent to which B.
hybridum genomes are a mixture of alleles derived from
B. stacei- and B. distachyon-like populations (Fig. 4). The
diversity of individual ancestries inferred for different B.
hybridum accessions (Fig. 4) is consistent with multiple
origins of this allotetraploid species, as proposed by the
research groups of P. Catalán and colleagues (López-Alvarez et al., 2012). The origination of these B. hybridum lines
from Portugal, Spain, Albania, and Iran further supports
the idea of multiple hybridization events. Strikingly, the
B. hybridum accessions surveyed here all carried ancestry from B. distachyon Population 2, perhaps indicating
a greater proclivity or opportunity for members of this
population to cross with B. stacei. It will be interesting to
determine whether a contribution from Population 2 is
consistently present when larger numbers of B. hybridum
accessions are examined.
The presence of only three B. stacei accessions in
our study reflects both the relative scarcity of established
B. stacei in Brachypodium collections, as well as our
intentional focus on B. distachyon as a model. The five
B. hybridum accessions newly identified in our study
were initially presumed to be B. distachyon based on
their morphology but clearly cluster with the reference
B. hybridum lines in principal component and STRUCTURE analyses (Fig. 3, 4). These results emphasize the
difficulty of making accurate species assignments for
individual Brachypodium accessions based solely on
morphological traits. While each species might, on average, be phenotypically distinct for a particular trait, the
intraspecific variation is large (Tyler et al., 2014), and the
phenotypic and geographic ranges can overlap (Mur et
al., 2011; Catalán et al., 2012).
With regard to the 84 accessions confirmed to be B.
distachyon, the data collected provide additional information about key reference lines and, importantly, allow
the characterization of new germplasm. Our success in
expanding the known genetic diversity for B. distachyon
is illustrated by our identification of over 54,000 SNPs
for accessions spanning the native range of this model
grass. This total is more than three times the number
of SNPs detected in a GBS analysis of Turkish accessions (Dell’Acqua et al., 2014). Interestingly, the inferred
ancestries of Bd21 and Bd21-3 contained a relatively high
the pl ant genome

ju ly 2016

vol . 9, no . 2
proportion of Population 2 in comparison to Bd3-1 and
most other early-flowering lines (Fig. 4; Supplemental
Data S6). This observation reinforces the utility of Bd3-1
as a mapping partner for crosses with Bd21 or Bd21-3.
The Bd21 and Bd3-1 lines have already been used to generate a recombinant inbred population, which has proven
to be a powerful tool in investigating the genetic basis of
traits such as pathogen resistance (Cui et al., 2012). Our
results could indicate additional promising candidates
for high-coverage genome resequencing, generating
recombinants, and other resource-intensive studies.
While Turkish inbred lines genotyped with 43 SSR
markers had separated into distinct phylogenetic clades
largely consistent with flowering time differences (Vogel
et al., 2009), our GBS data showed that this distinction
held across the native range of B. distachyon from Spain
in the west to Georgia in the east. In terms of evolutionary history, the structuring of B. distachyon into at least
two highly diverged genetic populations is intriguing,
especially because the two main populations occupy
similar geographic ranges and because the divergence
seems to be relatively recent. In some of our samples,
individuals belonging to different populations were
found in the same locality or close by. This result indicates there is little gene flow between populations, consistent with high levels of self-fertilization in B. distachyon
(Vogel et al., 2009). Although geographic subdivisions
are evident within our main populations, the broad
population structure is more strongly associated with
what appears as flowering time in our common garden
experiment. However, it is not intuitive why accessions
could have different flowering times and yet thrive in the
same locality subject to the same environmental conditions. We note, though, that the differences in flowering
time observed under our greenhouse conditions do not
necessarily represent true flowering times in the wild.
Accessions that flower at different times in our experiments may flower concurrently in natural populations,
despite divergent genetic backgrounds, as a response
to local environmental conditions that act as cues for
flowering. For example, the length of vernalization in
many localities may exceed the lengths of time tested in
our greenhouse. In fact, because most of our collections
were made at a single time point, it is possible that all
individuals at a single locality flower at similar times in
the wild, regardless of genetic background. Nevertheless,
our FST analyses strongly support the existence of some
reproductive-phase morphological or developmental
differences between the two main B. distachyon populations. Intriguingly, many genes close to SNPs highly
differentiated between populations seem to be involved
in the autonomous flowering pathway, suggesting that
observed flowering time differences might be related to
developmental phenotypes.
None of the fixed or high FST loci (Supplemental
Table S2) between populations overlapped with MTAs
detected by GWAS (Supplemental Table S4). This is
expected and suggests, first, that our tests for association
controlled well for confounding population structure
(Brachi et al., 2011) and, second, that phenotypic diversity between populations and within populations may
involve different flowering pathways. Polymorphic loci
underlying flowering differences between populations are
likely to display high FST and are best captured through
such population analyses. That our GWAS yielded nine
significant associations indicates that flowering time
variation within populations is still sufficient to permit
mapping of loci responsible for these differences. Additional investigation will be required to determine if any
casual polymorphisms exist within the five a priori candidate genes or potential novel regulators.
Whereas GWAS has recently been used in B. distachyon to identify genetic loci associated with environmental variables (Dell’Acqua et al., 2014), to our
knowledge, our findings are the first to illustrate the
feasibility of linking genotype to a developmental phenotype in B. distachyon. Examination of our MTAs for
known flowering-time, circadian-rhythm, and photoperiod genes revealed candidate genes underlying five
of the nine MTAs. Our identification of four MTAs far
from any previously described candidate genes suggests
that further study is needed to better understand regulation of flowering time in the grasses. Our estimates of
LD decay in B. distachyon indicate that improvements to
GWAS in this species can be achieved through expansion
of characterized germplasm, in particular from Population 1, which may decrease the size of genomic intervals
at which LD decays significantly.
Conclusions
As demonstrated for flowering time in B. distachyon,
GWAS can detect large effects even in small populations.
On the other hand, detecting significant MTAs of smalleffect or rare variants requires large populations (Atwell
et al., 2010). Further expanding B. distachyon germplasm
collections is therefore crucial for increased power in
future GWAS. Because flowering time is a major factor separating populations, harvesting multiple times
throughout the year at a single location could enhance
the diversity of the germplasm collected and increase
the likelihood that accessions most useful for GWAS—
early-flowering accessions from STRUCTURE-defined
Population 1—would be recovered. As well as capturing
additional diversity, the availability of new B. distachyon
accessions would increase the resolution of GWAS and
thus greatly facilitate the rapid identification of candidate
genes for functional testing in this model system.
Supplemental Material
Supplemental Tables S1–S4, Supplemental Data files S1–
S15 (including SNP genotypes in Supplemental Data S1
and S15), and Supplemental Fig. S1–S5 are available online:
Supplemental Table S1. Summary information for
accessions used in this study.
Supplemental Table S2. Annotation of genes near
highly divergent SNPs.
t yler et al .: b . distachyon popu l ations di ffer in flowering 17
of
20
Supplemental Table S3. Enrichment of GO terms for
genes near highly divergent SNPs.
Supplemental Table S4. Significant marker-trait association intervals as determined by mixed linear model
analysis for flowering time after 6 wk of vernalization.
Supplemental Data S1. High-quality SNPs for 94
Brachypodium accessions.
Supplemental Data S2. Number of SNPs compared to
the Bd21 reference.
Supplemental Data S3. Number of SNPs in pairwise
comparisons.
Supplemental Data S4. Principal component values
for 94 Brachypodium accessions.
Supplemental Data S5. Delta K results for 94 Brachypodium accessions.
Supplemental Data S6. Inferred individual ancestry
for 94 Brachypodium accessions.
Supplemental Data S7. Principal component values
for 84 B. distachyon accessions.
Supplemental Data S8. Delta K results for 84 B. distachyon accessions.
Supplemental Data S9. Inferred individual ancestry
for 84 B. distachyon accessions.
Supplemental Data S10. B. distachyon flowering
times after cold treatment.
Supplemental Data S11. LOSITAN results for highly
differentiated SNPs.
Supplemental Data S12. BayeScan results for highly
differentiated SNPs.
Supplemental Data S13. SNPs highly differentiated
between B. distachyon populations.
Supplemental Data S14. SNPs highly differentiated
between Brachypodium species.
Supplemental Data S15. High-quality SNPs for 84 B.
distachyon accessions.
Supplemental Fig. S1. Neighbor-joining tree for B.
distachyon.
Supplemental Fig. S2. AgriGo graphic of GO terms
for genes near highly divergent SNPs.
Supplemental Fig. S3. Timing of the B. distachyon
population split.
Supplemental Fig. S4. Quantile-Quantile plots for
general linear and mixed linear models for flowering time
of B. distachyon accessions after 6 wk of vernalization.
Supplemental Fig. S5. Box plots of flowering time following 6 wk of vernalization for each significant MTA
based on allele.
Acknowledgments
The authors thank the University of Oslo, Norway, for the Bioportal service used for STRUCTURE analysis. We are also grateful to Sherin Perera
and Zhongyun Huang (University of Massachusetts–Amherst) for technical assistance and help with a bioinformatic analysis, respectively. The
authors state they have no conflicts of interest to declare.
This material is based on work supported by the National Institute
of Food and Agriculture (NIFA), USDA, the Massachusetts Agricultural
Experiment Station, and the Biology Department of the University of
Massachusetts–Amherst under project number MAS00419 to A.L.C.
and S.P.H., and the Office of Science (BER) Department of Energy Grant
18
of
20
DE-SC0006621 to S.P.H. The contents are solely the responsibility of the
authors and do not necessarily represent the official views of the USDA,
NIFA, or other funding agencies.
References
Alves, S.C., B. Worland, V. Thole, J.W. Snape, M.W. Bevan, and P. Vain.
2009. A protocol for Agrobacterium-mediated transformation of
Brachypodium distachyon community standard line Bd21. Nat. Protoc.
4:638–649. doi:10.1038/nprot.2009.30
Antao, T., A. Lopes, R.J. Lopes, A. Beja-Pereira, and G. Luikart. 2008.
LOSITAN: A workbench to detect molecular adaptation based on a Fstoutlier method. BMC Bioinformatics 9:323. doi:10.1186/1471-2105-9-323
Atwell, S., Y.S. Huang, B.J. Vilhjálmsson, G. Willems, M. Horton, Y. Li,
et al. 2010. Genome-wide association study of 107 phenotypes in
Arabidopsis thaliana inbred lines. Nature 465:627–631. doi:10.1038/
nature08800
Beales, J., A. Turner, S. Griffiths, J.W. Snape, and D.A. Laurie. 2007. A
pseudo-response regulator is misexpressed in the photoperiod insensitive Ppd-D1a mutant of wheat (Triticum aestivum L.). Theor. Appl.
Genet. 115:721–733. doi:10.1007/s00122-007-0603-4
Beaumont, M.A., and R.A. Nichols. 1996. Evaluating loci for use in the
genetic analysis of population structure. Proc. R. Soc. Lond. B Biol. Sci.
263:1619–1626. doi:10.1098/rspb.1996.0237
Beaumont, M.A., W. Zhang, and D.J. Balding. 2002. Approximate Bayesian
computation in population genetics. Genetics 162:2025–2035.
Brachi, B., G.P. Morris, and J.O. Borevitz. 2011. Genome-wide association
studies in plants: The missing heritability is in the field. Genome Biol.
12:232. doi:10.1186/gb-2011-12-10-232
Bradbury, P.J., Z. Zhang, D.E. Kroon, T.M. Casstevens, Y. Ramdoss,
and E.S. Buckler. 2007. TASSEL: Software for association mapping
of complex traits in diverse samples. Bioinformatics 23:2633–2635.
doi:10.1093/bioinformatics/btm308
Bragg, J.N., L. Tyler, and J.P. Vogel. 2012. Brachypodium. In: C. Kole, C.P.
Joshi, and D.R. Shonnard, editors, Handbook of bioenergy crop plants.
Taylor and Francis, Boca Raton, FL. p. 593–618.
Brkljacic, J., E. Grotewold, R. Scholl, T. Mockler, D.F. Garvin, P. Vain, et al.
2011. Brachypodium as a model for the grasses: Today and the future.
Plant Physiol. 157:3–13. doi:10.1104/pp.111.179531
Camacho, C., G. Coulouris, V. Avagyan, N. Ma, J. Papadopoulos, K. Bealer,
and T.L. Madden. 2009. BLAST+: Architecture and applications. BMC
Bioinformatics 10:421. doi:10.1186/1471-2105-10-421
Catalán, P., J. Müller, R. Hasterok, G. Jenkins, L.A.J. Mur, T. Langdon, A.
Betekhtin, D. Siwinska, M. Pimentel, and D. López-Alvarez. 2012. Evolution and taxonomic split of the model grass Brachypodium distachyon.
Ann. Bot. (Lond.) 109:385–405. doi:10.1093/aob/mcr294
Cornuet, J.M., P. Pudlo, J. Veyssier, A. Dehne-Garcia, M. Gautier, R. Leblois,
J.M. Marin, and A. Estoup. 2014. DIYABC v2.0: A software to make
approximate Bayesian computation inferences about population history
using single nucleotide polymorphism, DNA sequence and microsatellite
data. Bioinformatics 30:1187–1189. doi:10.1093/bioinformatics/btt763
Cornuet, J.M., F. Santos, M.A. Beaumont, C.P. Robert, J.M. Marin, D.J.
Balding, T. Guillemaud, and A. Estoup. 2008. Inferring population history with DIY ABC: A user-friendly approach to approximate Bayesian
computation. Bioinformatics 24:2713–2719. doi:10.1093/bioinformatics/
btn514
Cui, Y., M.Y. Lee, N. Huo, J. Bragg, L. Yan, C. Yuan, C. Li, S.J. Holditch, J.
Xie, M.C. Luo, et al. 2012. Fine mapping of the Bsr1 barley stripe mosaic
virus resistance gene in the model grass Brachypodium distachyon.
PLoS ONE 7:E38333. doi:10.1371/journal.pone.0038333
Dell’Acqua, M., A. Zuccolo, M. Tuna, L. Gianfranceschi, and M.E. Pè. 2014.
Targeting environmental adaptation in the monocot model Brachypodium distachyon: A multi-faceted approach. BMC Genomics 15:801.
doi:10.1186/1471-2164-15-801
Ding, Z., and J. Friml. 2010. Auxin regulates distal stem cell differentiation in Arabidopsis roots. Proc. Natl. Acad. Sci. USA 107:12046–12051.
doi:10.1073/pnas.1000672107
Draper, J., L.A. Mur, G. Jenkins, G.C. Ghosh-Biswas, P. Bablak, R. Hasterok,
and A.P. Routledge. 2001. Brachypodium distachyon: A new model system for functional genomics in grasses. Plant Physiol. 127:1539–1555.
doi:10.1104/pp.010196
the pl ant genome

ju ly 2016

vol . 9, no . 2
Du, Z., X. Zhou, Y. Ling, Z. Zhang, and Z. Su. 2010. agriGO: A GO analysis
toolkit for the agricultural community. Nucleic Acids Res. 38:W64–W70.
doi:10.1093/nar/gkq310
Elshire, R.J., J.C. Glaubitz, Q. Sun, J.A. Poland, K. Kawamoto, E.S. Buckler,
and S.E. Mitchell. 2011. A robust, simple genotyping-by-sequencing
(GBS) approach for high diversity species. PLoS ONE 6:E19379.
doi:10.1371/journal.pone.0019379
Evanno, G., S. Regnaut, and J. Goudet. 2005. Detecting the number of
clusters of individuals using the software structure: A simulation study.
Mol. Ecol. 14:2611–2620. doi:10.1111/j.1365-294X.2005.02553.x
Falush, D., M. Stephens, and J.K. Pritchard. 2003. Inference of population
structure using multilocus genotype data: Linked loci and correlated
allele frequencies. Genetics 164:1567–1587.
Falush, D., M. Stephens, and J.K. Pritchard. 2007. Inference of population
structure using multilocus genotype data: Dominant markers and null
alleles. Mol. Ecol. Notes 7:574–578. doi:10.1111/j.1471-8286.2007.01758.x
Felsenstein, J. 1985. Confidence limits on phylogenies: An approach using
the bootstrap. Evolution 39:783–791. doi:10.2307/2408678
Filiz, E., B.S. Ozdemir, F. Budak, J.P. Vogel, M. Tuna, and H. Budak. 2009.
Molecular, morphological, and cytological analysis of diverse Brachypodium distachyon inbred lines. Genome 52:876–890. doi:10.1139/g09-062
Foll, M., and O. Gaggiotti. 2008. A genome-scan method to identify
selected loci appropriate for both dominant and codominant markers:
A Bayesian Perspective. Genetics 180:977–993. doi:10.1534/genetics.108.092221
Garvin, D.F., Y.Q. Gu, R. Hasterok, S.P. Hazen, G. Jenkins, T.C. Mockler,
L.A.J. Mur, and J.P. Vogel. 2008. Development of genetic and genomic
research resources for Brachypodium distachyon, a new model system for grass crop research RID B-3176-2009. Crop Sci. 48:S69–S84.
doi:10.2135/cropsci2007.06.0332tpg
Giraldo, P., M. Rodríguez-Quijano, J.F. Vázquez, J.M. Carrillo, and E.
Benavente. 2012. Validation of microsatellite markers for cytotype
discrimination in the model grass Brachypodium distachyon. Genome
55:523–527. doi:10.1139/g2012-039
Girin, T., L.C. David, C. Chardin, R. Sibout, A. Krapp, S. Ferrario-Méry,
and F. Daniel-Vedele. 2014. Brachypodium: A promising hub between
model species and cereals. J. Exp. Bot. 65:5683–5696. doi:10.1093/jxb/
eru376
Goff, S.A., M. Vaughn, S. McKay, E. Lyons, A.E. Stapleton, D. Gessler, N.
Matasci, L. Wang, M. Hanlon, A. Lenards, et al. 2011. The iPlant collaborative: Cyberinfrastructure for plant biology. Front. Plant Sci. 2:34.
Gomez, L.D., J.K. Bristow, E.R. Statham, and S.J. McQueen-Mason.
2008. Analysis of saccharification in Brachypodium distachyon
stems under mild conditions of hydrolysis. Biotechnol. Biofuels 1:15.
doi:10.1186/1754-6834-1-15
Gordon, S.P., H. Priest, D.L. Des Marais, W. Schackwitz, M. Figueroa, J.
Martin, et al. 2014. Genome diversity in Brachypodium distachyon:
Deep sequencing of highly diverse inbred lines. Plant J. 79:361–374.
doi:10.1111/tpj.12569
Greb, T., O. Clarenz, E. Schäfer, D. Müller, R. Herrero, G. Schmitz, and K.
Theres. 2003. Molecular analysis of the LATERAL SUPPRESSOR gene
in Arabidopsis reveals a conserved control mechanism for axillary
meristem formation. Genes Dev. 17:1175–1187. doi:10.1101/gad.260703
Gremski, K., G. Ditta, and M.F. Yanofsky. 2007. The HECATE genes regulate female reproductive tract development in Arabidopsis thaliana.
Development 134:3593–3601. doi:10.1242/dev.011510
Hasterok, R., J. Draper, and G. Jenkins. 2004. Laying the cytotaxonomic foundations of a new model grass, Brachypodium
distachyon (L.) Beauv. Chromosome Res. 12:397–403. doi:10.1023/
B:CHRO.0000034130.35983.99
Higgins, J.A., P.C. Bailey, and D.A. Laurie. 2010. Comparative genomics
of flowering time pathways using Brachypodium distachyon as a model
for the temperate grasses. PLoS ONE 5:E10065. doi:10.1371/journal.
pone.0010065
Hong, S.Y., J.H. Park, S.H. Cho, M.S. Yang, and C.M. Park. 2011. Phenological growth stages of Brachypodium distachyon: Codification and
description. Weed Res. 51:612–620. doi:10.1111/j.1365-3180.2011.00877.x
Huang, X., X. Wei, T. Sang, Q. Zhao, Q. Feng, Y. Zhao, C. Li, C. Zhu, T. Lu,
Z. Zhang, et al. 2010. Genome-wide association studies of 14 agronomic
traits in rice landraces. Nat. Genet. 42:961–967. doi:10.1038/ng.695
Hubisz, M.J., D. Falush, M. Stephens, and J.K. Pritchard. 2009. Inferring weak population structure with the assistance of sample group
information. Mol. Ecol. Resour. 9:1322–1332. doi:10.1111/j.17550998.2009.02591.x
Huson, D.H., D.C. Richter, C. Rausch, T. Dezulian, M. Franz, and R. Rupp.
2007. Dendroscope: An interactive viewer for large phylogenetic trees.
BMC Bioinformatics 8:460. doi:10.1186/1471-2105-8-460
Idziak, D., A. Betekhtin, E. Wolny, K. Lesniewska, J. Wright, M. Febrer,
M.W. Bevan, G. Jenkins, and R. Hasterok. 2011. Painting the chromosomes of Brachypodium-current status and future prospects. Chromosoma 120:469–479. doi:10.1007/s00412-011-0326-9
Ikeda, A., M. Ueguchi-Tanaka, Y. Sonoda, H. Kitano, M. Koshioka, Y.
Futsuhara, M. Matsuoka, and J. Yamaguchi. 2001. slender rice, a constitutive gibberellin response mutant, is caused by a null mutation of the
SLR1 gene, an ortholog of the height-regulating gene GAI/RGA/RHT/
D8. Plant Cell 13:999–1010. doi:10.1105/tpc.13.5.999
International Brachypodium Initiative. 2010. Genome sequencing and
analysis of the model grass Brachypodium distachyon. Nature 463:763–
768. doi:10.1038/nature08747
Jaeger, K.E., and P.A. Wigge. 2007. FT protein acts as a long-range signal in
Arabidopsis. Curr. Biol. 17:1050–1054. doi:10.1016/j.cub.2007.05.008
Jiang, G., Y. Xiang, J. Zhao, D. Yin, X. Zhao, L. Zhu, and W. Zhai. 2014.
Regulation of inflorescence branch development in rice through a novel
pathway involving the pentatricopeptide repeat protein sped1-D. Genetics 197:1395–1407. doi:10.1534/genetics.114.163931
Kardailsky, I., V.K. Shukla, J.H. Ahn, N. Dagenais, S.K. Christensen, J.T.
Nguyen, J. Chory, M.J. Harrison, and D. Weigel. 1999. Activation tagging of the floral inducer FT. Science 286:1962–1965. doi:10.1126/science.286.5446.1962
Kim, S., V. Plagnol, T.T. Hu, C. Toomajian, R.M. Clark, S. Ossowski, J.R.
Ecker, D. Weigel, and M. Nordborg. 2007. Recombination and linkage disequilibrium in Arabidopsis thaliana. Nat. Genet. 39:1151–1155.
doi:10.1038/ng2115
Konieczny, A., and F.M. Ausubel. 1993. A procedure for mapping Arabidopsis mutations using co-dominant ecotype-specific PCR-based markers.
Plant J. 4:403–410. doi:10.1046/j.1365-313X.1993.04020403.x
Koornneef, M., and D. Meinke. 2010. The development of Arabidopsis as
a model plant. Plant J. 61:909–921. doi:10.1111/j.1365-313X.2009.04086.x
Lee, S.J., T.A. Warnick, S. Pattathil, J.G. Alvelo-Maurosa, M.J. Serapiglia, H.
McCormick, et al. 2012. Biological conversion assay using Clostridium
phytofermentans to estimate plant feedstock quality. Biotechnol. Biofuels 5:5. doi:10.1186/1754-6834-5-5
Li, J., Y. Li, S. Chen, and L. An. 2010. Involvement of brassinosteroid signals
in the floral-induction network of Arabidopsis. J. Exp. Bot. doi:10.1093/
jxb/erq241
López-Alvarez, D., M.L. López-Herranz, A. Betekhtin, and P. Catalán. 2012.
A DNA barcoding method to discriminate between the model plant
Brachypodium distachyon and its close relatives B. stacei and B. hybridum (Poaceae). PLoS ONE 7:E51058. doi:10.1371/journal.pone.0051058
López-Alvarez, D., A.J. Manzaneda, P.J. Rey, P. Giraldo, E. Benavente, J.
Allainguillaume, et al. 2015. Environmental niche variation and evolutionary diversification of the Brachypodium distachyon grass complex species in their native circum-Mediterranean range. Am. J. Bot.
102:1073–1088. doi:10.3732/ajb.1500128
Maier, A.T., S. Stehling-Sun, S.L. Offenburger, and J.U. Lohmann. 2011. The
bZIP transcription factor PERIANTHIA: A multifunctional hub for
meristem control. Front. Plant Sci. 2:79. doi:10.3389/fpls.2011.00079
Malcomber, S.T., and E.A. Kellogg. 2005. SEPALLATA gene diversification: Brave new whorls. Trends Plant Sci. 10:427–435. doi:10.1016/j.
tplants.2005.07.008
Marriott, P.E., R. Sibout, C. Lapierre, J.U. Fangel, W.G.T. Willats, H.
Hofte, L.D. Gómez, and S.J. McQueen-Mason. 2014. Range of cellwall alterations enhance saccharification in Brachypodium distachyon
mutants. Proc. Natl. Acad. Sci. USA 111: 14601–14606. doi:10.1073/
pnas.1414020111
Mather, K.A., A.L. Caicedo, N.R. Polato, K.M. Olsen, S. McCouch, and
M.D. Purugganan. 2007. The extent of linkage disequilibrium in
rice (Oryza sativa L.). Genetics 177:2223–2232. doi:10.1534/genetics.107.079616
t yler et al .: b . distachyon popu l ations di ffer in flowering 19
of
20
Mochida, K., and K. Shinozaki. 2013. Unlocking Triticeae genomics to sustainably feed the future. Plant Cell Physiol. 54:1931–1950. doi:10.1093/
pcp/pct163
Mosquna, A., A. Katz, E.L. Decker, S.A. Rensing, R. Reski, and N. Ohad.
2009. Regulation of stem cell maintenance by the Polycomb protein
FIE has been conserved during land plant evolution. Development
136:2433–2444. doi:10.1242/dev.035048
Mur, L.A.J., J. Allainguillaume, P. Catalán, R. Hasterok, G. Jenkins, K.
Lesniewska, I. Thomas, and J. Vogel. 2011. Exploiting the Brachypodium tool box in cereal and grass research. New Phytol. 191:334–347.
doi:10.1111/j.1469-8137.2011.03748.x
Nakamichi, N., M. Kita, S. Ito, T. Yamashino, and T. Mizuno. 2005.
PSEUDO-RESPONSE REGULATORS, PRR9, PRR7 and PRR5,
together play essential roles close to the circadian clock of Arabidopsis
thaliana. Plant Cell Physiol. 46:686–698. doi:10.1093/pcp/pci086
Nei, M., and S. Kumar. 2000. Molecular evolution and phylogenetics.
Oxford Univ. Press, New York.
Nole-Wilson, S., and B.A. Krizek. 2006. AINTEGUMENTA contributes
to organ polarity and regulates growth of lateral organs in combination with YABBY Genes. Plant Physiol. 141:977–987. doi:10.1104/
pp.106.076604
Pabón-Mora, N., G.K.S. Wong, and B.A. Ambrose. 2014. Evolution of fruit
development genes in flowering plants. Front. Plant Sci. 5:300.
Pajoro, A., P. Madrigal, J.M. Muiño, J.T. Matus, J. Jin, M.A. Mecchia, et
al. 2014. Dynamics of chromatin accessibility and gene regulation by
MADS-domain transcription factors in flower development. Genome
Biol. 15:R41. doi:10.1186/gb-2014-15-3-r41
Pasam, R.K., R. Sharma, M. Malosetti, F.A. van Eeuwijk, G. Haseneyer, B.
Kilian, and A. Graner. 2012. Genome-wide association studies for agronomical traits in a world wide spring barley collection. BMC Plant Biol.
12:16. doi:10.1186/1471-2229-12-16
Patterson, N., A.L. Price, and D. Reich. 2006. Population structure and
eigenanalysis. PLoS Genet. 2:E190. doi:10.1371/journal.pgen.0020190
Price, A.L., N.J. Patterson, R.M. Plenge, M.E. Weinblatt, N.A. Shadick,
and D. Reich. 2006. Principal components analysis corrects for stratification in genome-wide association studies. Nat. Genet. 38:904–909.
doi:10.1038/ng1847
Pritchard, J.K., M. Stephens, and P. Donnelly. 2000. Inference of population
structure using multilocus genotype data. Genetics 155:945–959.
R Development Core Team. 2014. R: A language and environment for statistical computing. R Foundation for Stat. Comput., Vienna, Austria.
Ream, T.S., D.P. Woods, C.J. Schwartz, C.P. Sanabria, J.A. Mahoy, E.M.
Walters, H.F. Kaeppler, and R.M. Amasino. 2014. Interaction of photoperiod and vernalization determines flowering time of Brachypodium
distachyon. Plant Physiol. 164:694–709. doi:10.1104/pp.113.232678
Saitou, N., and M. Nei. 1987. The neighbor-joining method: A new method
for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406–425.
Samach, A., H. Onouchi, S.E. Gold, G.S. Ditta, Z. Schwarz-Sommer, M.F.
Yanofsky, and G. Coupland. 2000. Distinct roles of CONSTANS target
genes in reproductive development of Arabidopsis. Science 288:1613–
1616. doi:10.1126/science.288.5471.1613
Schlereth, A., B. Möller, W. Liu, M. Kientz, J. Flipse, E.H. Rademacher, M.
Schmid, G. Jürgens, and D. Weijers. 2010. MONOPTEROS controls
embryonic root initiation by regulating a mobile transcription factor.
Nature 464:913–916. doi:10.1038/nature08836
Schulze, S., B.N. Schäfer, E.A. Parizotto, O. Voinnet, and K. Theres. 2010.
LOST MERISTEMS genes regulate cell differentiation of central zone
descendants in Arabidopsis shoot meristems. Plant J. 64:668–678.
doi:10.1111/j.1365-313X.2010.04359.x
Schwartz, C.J., M.R. Doyle, A.J. Manzaneda, P.J. Rey, T. Mitchell-Olds, and
R.M. Amasino. 2010. Natural variation of flowering time and vernalization responsiveness in Brachypodium distachyon. BioEnergy Res.
3:38–46. doi:10.1007/s12155-009-9069-3
Schwarz, S., A.V. Grande, N. Bujdoso, H. Saedler, and P. Huijser. 2008. The
microRNA regulated SBP-box genes SPL9 and SPL15 control shoot
maturation in Arabidopsis. Plant Mol. Biol. 67:183–195. doi:10.1007/
s11103-008-9310-z
20
of
20
Slavov, G.T., R. Nipper, P. Robson, K. Farrar, G.G. Allison, M. Bosch, J.C.
Clifton-Brown, I.S. Donnison, and E. Jensen. 2014. Genome-wide association studies and prediction of 17 traits related to phenology, biomass
and cell wall composition in the energy grass Miscanthus sinensis. New
Phytol. 201:1227–1239. doi:10.1111/nph.12621
Stamatakis, A. 2014. RAxML version 8: A tool for phylogenetic analysis
and post-analysis of large phylogenies. Bioinformatics 30:1312–1313.
doi:10.1093/bioinformatics/btu033
Taberlet, P., L. Gielly, G. Pautou, and J. Bouvet. 1991. Universal primers for
amplification of three non-coding regions of chloroplast DNA. Plant
Mol. Biol. 17:1105–1109. doi:10.1007/BF00037152
Tamura, K., D. Peterson, N. Peterson, G. Stecher, M. Nei, and S. Kumar.
2011. MEGA5: Molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony
methods. Mol. Biol. Evol. 28:2731–2739. doi:10.1093/molbev/msr121
Tian, F., P.J. Bradbury, P.J. Brown, H. Hung, Q. Sun, S. Flint-Garcia, T.R.
Rocheford, M.D. McMullen, J.B. Holland, and E.S. Buckler. 2011.
Genome-wide association study of leaf architecture in the maize
nested association mapping population. Nat. Genet. 43:159–162.
doi:10.1038/ng.746
Tyler, L., J.U. Fangel, A.D. Fagerström, M.A. Steinwand, T.K. Raab, W.G.
Willats, and J.P. Vogel. 2014. Selection and phenotypic characterization of a core collection of Brachypodium distachyon inbred lines. BMC
Plant Biol. 14:25. doi:10.1186/1471-2229-14-25
Vain, P., B. Worland, V. Thole, N. McKenzie, S.C. Alves, M. Opanowicz,
L.J. Fish, M.W. Bevan, and J.W. Snape. 2008. Agrobacterium-mediated
transformation of the temperate grass Brachypodium distachyon (genotype Bd21) for T-DNA insertional mutagenesis. Plant Biotechnol. J.
6:236–245. doi:10.1111/j.1467-7652.2007.00308.x
Vogel, J.P., D.F. Garvin, O.M. Leong, and D.M. Hayden. 2006. Agrobacterium-mediated transformation and inbred line development in the
model grass Brachypodium distachyon. Plant Cell Tissue Organ Cult.
84:199–211. doi:10.1007/s11240-005-9023-9
Vogel, J., and T. Hill. 2008. High-efficiency Agrobacterium-mediated transformation of Brachypodium distachyon inbred line Bd21-3. Plant Cell
Rep. 27:471–478. doi:10.1007/s00299-007-0472-y
Vogel, J.P., M. Tuna, H. Budak, N. Huo, Y.Q. Gu, and M.A. Steinwand.
2009. Development of SSR markers and analysis of diversity in Turkish populations of Brachypodium distachyon. BMC Plant Biol. 9:88.
doi:10.1186/1471-2229-9-88
Wickham, H. (2009) ggplot2: Elegant graphics for data analysis. Springer
Science & Business Media, New York.
Xu, X., X. Liu, S. Ge, J.D. Jensen, F. Hu, X. Li, et al. 2012. Resequencing
50 accessions of cultivated and wild rice yields markers for identifying agronomically important genes. Nat. Biotechnol. 30:105–111.
doi:10.1038/nbt.2050
Yan, J., T. Shah, M.L. Warburton, E.S. Buckler, M.D. McMullen, and J.
Crouch. 2009. Genetic characterization and linkage disequilibrium
estimation of a global maize collection using SNP markers. PLoS ONE
4:E8451. doi:10.1371/journal.pone.0008451
Yano, M., Y. Katayose, M. Ashikari, U. Yamanouchi, L. Monna, T. Fuse, et
al. 2000. Hd1, a major photoperiod sensitivity quantitative trait locus
in rice, is closely related to the Arabidopsis flowering time gene CONSTANS. Plant Cell 12:2473–2483. doi:10.1105/tpc.12.12.2473
Yu, J., G. Pressoir, W.H. Briggs, I. Vroh Bi, M. Yamasaki, J.F. Doebley, et
al. 2006. A unified mixed-model method for association mapping that
accounts for multiple levels of relatedness. Nat. Genet. 38:203–208.
doi:10.1038/ng1702
Zagotta, M., S. Shannon, C. Jacobs, and D. Meeks-Wagner. 1992. Early-flowering mutants of Arabidopsis thaliana. Funct. Plant Biol. 19:411–418.
Zakhrabekova, S., S.P. Gough, I. Braumann, A.H. Müller, J. Lundqvist, K.
Ahmann, et al. 2012. Induced mutations in circadian clock regulator
Mat-a facilitated short-season adaptation and range extension in cultivated barley. Proc. Natl. Acad. Sci. USA 109:4326–4331. doi:10.1073/
pnas.1113009109
Zhang, Z., E. Ersoz, C.Q. Lai, R.J. Todhunter, H.K. Tiwari, M.A. Gore, et al.
2010. Mixed linear model approach adapted for genome-wide association studies. Nat. Genet. 42:355–360. doi:10.1038/ng.546
the pl ant genome

ju ly 2016

vol . 9, no . 2